Warning: Permanently added '2620:52:3:1:dead:beef:cafe:c1d7' (ED25519) to the list of known hosts.

You can reproduce this build on your computer by running:

  sudo dnf install copr-rpmbuild
  /usr/bin/copr-rpmbuild --verbose --drop-resultdir --task-url https://copr.fedorainfracloud.org/backend/get-build-task/8809063-fedora-rawhide-x86_64 --chroot fedora-rawhide-x86_64


Version: 1.2
PID: 34127
Logging PID: 34128
Task:
{'allow_user_ssh': False,
 'appstream': False,
 'background': True,
 'build_id': 8809063,
 'buildroot_pkgs': [],
 'chroot': 'fedora-rawhide-x86_64',
 'enable_net': False,
 'fedora_review': False,
 'git_hash': '4be534733cb590e49c1555fa12730a07a5453460',
 'git_repo': 'https://copr-dist-git.fedorainfracloud.org/git/lecris/cmake-4.0/llama-cpp',
 'isolation': 'default',
 'memory_reqs': 2048,
 'package_name': 'llama-cpp',
 'package_version': 'b4580-2',
 'project_dirname': 'cmake-4.0',
 'project_name': 'cmake-4.0',
 'project_owner': 'lecris',
 'repo_priority': None,
 'repos': [{'baseurl': 'https://download.copr.fedorainfracloud.org/results/lecris/cmake-4.0/fedora-rawhide-x86_64/',
            'id': 'copr_base',
            'name': 'Copr repository',
            'priority': None}],
 'sandbox': 'lecris/cmake-4.0--lecris',
 'source_json': {},
 'source_type': None,
 'ssh_public_keys': None,
 'storage': 0,
 'submitter': 'lecris',
 'tags': [],
 'task_id': '8809063-fedora-rawhide-x86_64',
 'timeout': 115200,
 'uses_devel_repo': False,
 'with_opts': [],
 'without_opts': []}

Running: git clone https://copr-dist-git.fedorainfracloud.org/git/lecris/cmake-4.0/llama-cpp /var/lib/copr-rpmbuild/workspace/workdir-s56ti4b2/llama-cpp --depth 500 --no-single-branch --recursive

cmd: ['git', 'clone', 'https://copr-dist-git.fedorainfracloud.org/git/lecris/cmake-4.0/llama-cpp', '/var/lib/copr-rpmbuild/workspace/workdir-s56ti4b2/llama-cpp', '--depth', '500', '--no-single-branch', '--recursive']
cwd: .
rc: 0
stdout: 
stderr: Cloning into '/var/lib/copr-rpmbuild/workspace/workdir-s56ti4b2/llama-cpp'...

Running: git checkout 4be534733cb590e49c1555fa12730a07a5453460 --

cmd: ['git', 'checkout', '4be534733cb590e49c1555fa12730a07a5453460', '--']
cwd: /var/lib/copr-rpmbuild/workspace/workdir-s56ti4b2/llama-cpp
rc: 0
stdout: 
stderr: Note: switching to '4be534733cb590e49c1555fa12730a07a5453460'.

You are in 'detached HEAD' state. You can look around, make experimental
changes and commit them, and you can discard any commits you make in this
state without impacting any branches by switching back to a branch.

If you want to create a new branch to retain commits you create, you may
do so (now or later) by using -c with the switch command. Example:

  git switch -c <new-branch-name>

Or undo this operation with:

  git switch -

Turn off this advice by setting config variable advice.detachedHead to false

HEAD is now at 4be5347 automatic import of llama-cpp

Running: dist-git-client sources
INFO: Reading stdout from command: md5sum llama.cpp-b4580.tar.gz

/usr/bin/tail: /var/lib/copr-rpmbuild/main.log: file truncated
Running (timeout=115200): unbuffer mock --spec /var/lib/copr-rpmbuild/workspace/workdir-s56ti4b2/llama-cpp/llama-cpp.spec --sources /var/lib/copr-rpmbuild/workspace/workdir-s56ti4b2/llama-cpp --resultdir /var/lib/copr-rpmbuild/results --uniqueext 1742774257.650632 -r /var/lib/copr-rpmbuild/results/configs/child.cfg
INFO: mock.py version 6.1 starting (python version = 3.13.0, NVR = mock-6.1-1.fc41), args: /usr/libexec/mock/mock --spec /var/lib/copr-rpmbuild/workspace/workdir-s56ti4b2/llama-cpp/llama-cpp.spec --sources /var/lib/copr-rpmbuild/workspace/workdir-s56ti4b2/llama-cpp --resultdir /var/lib/copr-rpmbuild/results --uniqueext 1742774257.650632 -r /var/lib/copr-rpmbuild/results/configs/child.cfg
Start(bootstrap): init plugins
INFO: tmpfs initialized
INFO: selinux enabled
INFO: chroot_scan: initialized
INFO: compress_logs: initialized
Finish(bootstrap): init plugins
Start: init plugins
INFO: tmpfs initialized
INFO: selinux enabled
INFO: chroot_scan: initialized
INFO: compress_logs: initialized
Finish: init plugins
INFO: Signal handler active
Start: run
INFO: Start(/var/lib/copr-rpmbuild/workspace/workdir-s56ti4b2/llama-cpp/llama-cpp.spec)  Config(fedora-rawhide-x86_64)
Start: clean chroot
Finish: clean chroot
Mock Version: 6.1
INFO: Mock Version: 6.1
Start(bootstrap): chroot init
INFO: mounting tmpfs at /var/lib/mock/fedora-rawhide-x86_64-bootstrap-1742774257.650632/root.
INFO: calling preinit hooks
INFO: enabled root cache
INFO: enabled package manager cache
Start(bootstrap): cleaning package manager metadata
Finish(bootstrap): cleaning package manager metadata
INFO: Guessed host environment type: unknown
INFO: Using container image: registry.fedoraproject.org/fedora:rawhide
INFO: Pulling image: registry.fedoraproject.org/fedora:rawhide
INFO: Tagging container image as mock-bootstrap-dc3e80c1-e115-4b91-88c4-3863f04953cb
INFO: Checking that 12857a8ac9b526c53df07148d4c54be23787b288655ceb4ed5ca0227cd97623c image matches host's architecture
INFO: Copy content of container 12857a8ac9b526c53df07148d4c54be23787b288655ceb4ed5ca0227cd97623c to /var/lib/mock/fedora-rawhide-x86_64-bootstrap-1742774257.650632/root
INFO: mounting 12857a8ac9b526c53df07148d4c54be23787b288655ceb4ed5ca0227cd97623c with podman image mount
INFO: image 12857a8ac9b526c53df07148d4c54be23787b288655ceb4ed5ca0227cd97623c as /var/lib/containers/storage/overlay/3cf7b1887b1397f565d61fb9e6c6426872e2a8def215358985d497675867a245/merged
INFO: umounting image 12857a8ac9b526c53df07148d4c54be23787b288655ceb4ed5ca0227cd97623c (/var/lib/containers/storage/overlay/3cf7b1887b1397f565d61fb9e6c6426872e2a8def215358985d497675867a245/merged) with podman image umount
INFO: Removing image mock-bootstrap-dc3e80c1-e115-4b91-88c4-3863f04953cb
INFO: Package manager dnf5 detected and used (fallback)
INFO: Not updating bootstrap chroot, bootstrap_image_ready=True
Start(bootstrap): creating root cache
Finish(bootstrap): creating root cache
Finish(bootstrap): chroot init
Start: chroot init
INFO: mounting tmpfs at /var/lib/mock/fedora-rawhide-x86_64-1742774257.650632/root.
INFO: calling preinit hooks
INFO: enabled root cache
INFO: enabled package manager cache
Start: cleaning package manager metadata
Finish: cleaning package manager metadata
INFO: enabled HW Info plugin
INFO: Package manager dnf5 detected and used (direct choice)
INFO: Buildroot is handled by package management downloaded with a bootstrap image:
  rpm-4.20.1-1.fc43.x86_64
  rpm-sequoia-1.7.0-5.fc43.x86_64
  dnf5-5.2.12.0-1.fc43.x86_64
  dnf5-plugins-5.2.12.0-1.fc43.x86_64
Start: installing minimal buildroot with dnf5
Updating and loading repositories:
 fedora                                 100% | 546.6 KiB/s |  14.2 KiB |  00m00s
 Copr repository                        100% |  46.5 KiB/s |   1.5 KiB |  00m00s
 Copr repository                        100% |  15.0 MiB/s |   1.6 MiB |  00m00s
Repositories loaded.
Package                            Arch   Version                    Repository      Size
Installing group/module packages:
 bash                              x86_64 5.2.37-3.fc43              fedora       8.2 MiB
 bzip2                             x86_64 1.0.8-20.fc42              fedora      99.3 KiB
 coreutils                         x86_64 9.6-2.fc43                 fedora       5.4 MiB
 cpio                              x86_64 2.15-2.fc41                fedora       1.1 MiB
 diffutils                         x86_64 3.10-9.fc42                fedora       1.6 MiB
 fedora-release-common             noarch 43-0.7                     fedora      20.3 KiB
 findutils                         x86_64 1:4.10.0-5.fc42            fedora       1.9 MiB
 gawk                              x86_64 5.3.1-1.fc42               fedora       1.7 MiB
 glibc-minimal-langpack            x86_64 2.41.9000-2.fc43           fedora       0.0   B
 grep                              x86_64 3.11-10.fc42               fedora       1.0 MiB
 gzip                              x86_64 1.13-3.fc42                fedora     392.9 KiB
 info                              x86_64 7.2-3.fc42                 fedora     357.9 KiB
 patch                             x86_64 2.7.6-26.fc42              fedora     258.7 KiB
 redhat-rpm-config                 noarch 342-2.fc42                 fedora     186.8 KiB
 rpm-build                         x86_64 4.20.1-1.fc43              fedora     168.7 KiB
 sed                               x86_64 4.9-4.fc42                 fedora     857.3 KiB
 shadow-utils                      x86_64 2:4.17.4-1.fc43            fedora       4.0 MiB
 tar                               x86_64 2:1.35-5.fc42              fedora       3.0 MiB
 unzip                             x86_64 6.0-66.fc42                fedora     390.3 KiB
 util-linux                        x86_64 2.40.4-7.fc43              fedora       3.4 MiB
 which                             x86_64 2.23-1.fc42                fedora      83.4 KiB
 xz                                x86_64 1:5.6.3-3.fc42             fedora       1.2 MiB
Installing dependencies:
 add-determinism                   x86_64 0.6.0-1.fc43               fedora       2.5 MiB
 alternatives                      x86_64 1.32-1.fc43                fedora      62.2 KiB
 ansible-srpm-macros               noarch 1-17.1.fc42                fedora      35.7 KiB
 audit-libs                        x86_64 4.0.3-2.fc42               fedora     351.3 KiB
 binutils                          x86_64 2.44-3.fc43                fedora      25.9 MiB
 build-reproducibility-srpm-macros noarch 0.6.0-1.fc43               fedora     735.0   B
 bzip2-libs                        x86_64 1.0.8-20.fc42              fedora      84.6 KiB
 ca-certificates                   noarch 2024.2.69_v8.0.401-5.fc42  fedora       2.6 MiB
 coreutils-common                  x86_64 9.6-2.fc43                 fedora      11.1 MiB
 crypto-policies                   noarch 20250305-1.gita35b0fa.fc43 fedora     136.4 KiB
 curl                              x86_64 8.13.0~rc2-1.fc43          fedora     457.0 KiB
 cyrus-sasl-lib                    x86_64 2.1.28-30.fc42             fedora       2.3 MiB
 debugedit                         x86_64 5.1-5.fc43                 fedora     192.7 KiB
 dwz                               x86_64 0.15-9.fc42                fedora     291.0 KiB
 ed                                x86_64 1.21-2.fc42                fedora     146.5 KiB
 efi-srpm-macros                   noarch 6-3.fc43                   fedora      40.1 KiB
 elfutils                          x86_64 0.192-8.fc42               fedora       2.7 MiB
 elfutils-debuginfod-client        x86_64 0.192-8.fc42               fedora      83.9 KiB
 elfutils-default-yama-scope       noarch 0.192-8.fc42               fedora       1.8 KiB
 elfutils-libelf                   x86_64 0.192-8.fc42               fedora       1.2 MiB
 elfutils-libs                     x86_64 0.192-8.fc42               fedora     675.0 KiB
 fedora-gpg-keys                   noarch 43-0.1                     fedora     128.2 KiB
 fedora-release                    noarch 43-0.7                     fedora       0.0   B
 fedora-release-identity-basic     noarch 43-0.7                     fedora     719.0   B
 fedora-repos                      noarch 43-0.1                     fedora       4.9 KiB
 fedora-repos-rawhide              noarch 43-0.1                     fedora       2.2 KiB
 file                              x86_64 5.46-1.fc42                fedora     100.2 KiB
 file-libs                         x86_64 5.46-1.fc42                fedora      11.9 MiB
 filesystem                        x86_64 3.18-38.fc43               fedora     112.0   B
 filesystem-srpm-macros            noarch 3.18-38.fc43               fedora      38.2 KiB
 fonts-srpm-macros                 noarch 1:2.0.5-21.fc42            fedora      55.8 KiB
 forge-srpm-macros                 noarch 0.4.0-2.fc42               fedora      38.9 KiB
 fpc-srpm-macros                   noarch 1.3-14.fc42                fedora     144.0   B
 gdb-minimal                       x86_64 16.2-1.fc43                fedora      13.3 MiB
 gdbm-libs                         x86_64 1:1.23-9.fc42              fedora     129.9 KiB
 ghc-srpm-macros                   noarch 1.9.2-2.fc42               fedora     779.0   B
 glibc                             x86_64 2.41.9000-2.fc43           fedora       6.7 MiB
 glibc-common                      x86_64 2.41.9000-2.fc43           fedora       1.0 MiB
 glibc-gconv-extra                 x86_64 2.41.9000-2.fc43           fedora       7.2 MiB
 gmp                               x86_64 1:6.3.0-3.fc43             fedora     819.2 KiB
 gnat-srpm-macros                  noarch 6-7.fc42                   fedora       1.0 KiB
 go-srpm-macros                    noarch 3.6.0-6.fc42               fedora      60.8 KiB
 jansson                           x86_64 2.14-2.fc42                fedora      93.1 KiB
 json-c                            x86_64 0.18-2.fc42                fedora      86.7 KiB
 kernel-srpm-macros                noarch 1.0-25.fc42                fedora       1.9 KiB
 keyutils-libs                     x86_64 1.6.3-5.fc42               fedora      58.3 KiB
 krb5-libs                         x86_64 1.21.3-5.fc42              fedora       2.3 MiB
 libacl                            x86_64 2.3.2-3.fc42               fedora      38.3 KiB
 libarchive                        x86_64 3.7.7-4.fc43               fedora     930.6 KiB
 libattr                           x86_64 2.5.2-5.fc42               fedora      27.1 KiB
 libblkid                          x86_64 2.40.4-7.fc43              fedora     262.4 KiB
 libbrotli                         x86_64 1.1.0-6.fc43               copr_base  829.3 KiB
 libcap                            x86_64 2.73-2.fc42                fedora     207.1 KiB
 libcap-ng                         x86_64 0.8.5-4.fc42               fedora      72.9 KiB
 libcom_err                        x86_64 1.47.2-3.fc42              fedora      67.1 KiB
 libcurl                           x86_64 8.13.0~rc2-1.fc43          fedora     870.4 KiB
 libeconf                          x86_64 0.7.6-1.fc43               copr_base   64.6 KiB
 libevent                          x86_64 2.1.12-15.fc42             fedora     903.1 KiB
 libfdisk                          x86_64 2.40.4-7.fc43              fedora     372.3 KiB
 libffi                            x86_64 3.4.7-2.fc43               fedora      82.6 KiB
 libgcc                            x86_64 15.0.1-0.10.fc43           fedora     266.6 KiB
 libgomp                           x86_64 15.0.1-0.10.fc43           fedora     535.9 KiB
 libidn2                           x86_64 2.3.8-1.fc43               fedora     552.5 KiB
 libmount                          x86_64 2.40.4-7.fc43              fedora     356.2 KiB
 libnghttp2                        x86_64 1.65.0-1.fc43              fedora     162.2 KiB
 libpkgconf                        x86_64 2.3.0-2.fc42               fedora      78.1 KiB
 libpsl                            x86_64 0.21.5-5.fc42              fedora      76.4 KiB
 libselinux                        x86_64 3.8-1.fc42                 fedora     193.1 KiB
 libsemanage                       x86_64 3.8-1.fc42                 fedora     308.4 KiB
 libsepol                          x86_64 3.8-1.fc42                 fedora     826.0 KiB
 libsmartcols                      x86_64 2.40.4-7.fc43              fedora     180.4 KiB
 libssh                            x86_64 0.11.1-4.fc42              fedora     565.5 KiB
 libssh-config                     noarch 0.11.1-4.fc42              fedora     277.0   B
 libstdc++                         x86_64 15.0.1-0.10.fc43           fedora       2.8 MiB
 libtasn1                          x86_64 4.20.0-1.fc43              fedora     176.3 KiB
 libtool-ltdl                      x86_64 2.5.4-4.fc42               fedora      70.1 KiB
 libunistring                      x86_64 1.1-9.fc42                 fedora       1.7 MiB
 libuuid                           x86_64 2.40.4-7.fc43              fedora      37.3 KiB
 libverto                          x86_64 0.3.2-10.fc42              fedora      25.4 KiB
 libxcrypt                         x86_64 4.4.38-6.fc43              fedora     284.5 KiB
 libxml2                           x86_64 2.12.10-1.fc43             fedora       1.7 MiB
 libzstd                           x86_64 1.5.7-1.fc43               fedora     807.8 KiB
 lua-libs                          x86_64 5.4.7-3.fc43               fedora     276.9 KiB
 lua-srpm-macros                   noarch 1-15.fc42                  fedora       1.3 KiB
 lz4-libs                          x86_64 1.10.0-2.fc42              fedora     157.4 KiB
 mpfr                              x86_64 4.2.2-1.fc43               fedora     828.8 KiB
 ncurses-base                      noarch 6.5-5.20250125.fc42        fedora     326.8 KiB
 ncurses-libs                      x86_64 6.5-5.20250125.fc42        fedora     946.3 KiB
 ocaml-srpm-macros                 noarch 10-4.fc42                  fedora       1.9 KiB
 openblas-srpm-macros              noarch 2-19.fc42                  fedora     112.0   B
 openldap                          x86_64 2.6.9-3.fc42               fedora     655.1 KiB
 openssl-libs                      x86_64 1:3.5.0-1.fc43             fedora       8.9 MiB
 p11-kit                           x86_64 0.25.5-5.fc42              fedora       2.2 MiB
 p11-kit-trust                     x86_64 0.25.5-5.fc42              fedora     395.5 KiB
 package-notes-srpm-macros         noarch 0.5-13.fc42                fedora       1.6 KiB
 pam-libs                          x86_64 1.7.0-4.fc42               fedora     126.7 KiB
 pcre2                             x86_64 10.45-1.fc43               fedora     697.7 KiB
 pcre2-syntax                      noarch 10.45-1.fc43               fedora     273.9 KiB
 perl-srpm-macros                  noarch 1-57.fc42                  fedora     861.0   B
 pkgconf                           x86_64 2.3.0-2.fc42               fedora      88.5 KiB
 pkgconf-m4                        noarch 2.3.0-2.fc42               fedora      14.4 KiB
 pkgconf-pkg-config                x86_64 2.3.0-2.fc42               fedora     989.0   B
 popt                              x86_64 1.19-8.fc42                fedora     132.8 KiB
 publicsuffix-list-dafsa           noarch 20250116-1.fc42            fedora      68.5 KiB
 pyproject-srpm-macros             noarch 1.18.1-1.fc43              fedora       1.9 KiB
 python-srpm-macros                noarch 3.13-4.fc42                fedora      51.0 KiB
 qt5-srpm-macros                   noarch 5.15.16-1.fc43             fedora     500.0   B
 qt6-srpm-macros                   noarch 6.8.2-2.fc43               fedora     464.0   B
 readline                          x86_64 8.2-13.fc43                fedora     485.0 KiB
 rpm                               x86_64 4.20.1-1.fc43              fedora       3.1 MiB
 rpm-build-libs                    x86_64 4.20.1-1.fc43              fedora     206.6 KiB
 rpm-libs                          x86_64 4.20.1-1.fc43              fedora     721.8 KiB
 rpm-sequoia                       x86_64 1.7.0-5.fc43               fedora       2.4 MiB
 rust-srpm-macros                  noarch 26.3-4.fc42                fedora       4.8 KiB
 setup                             noarch 2.15.0-23.fc43             fedora     724.7 KiB
 sqlite-libs                       x86_64 3.49.0-1.fc43              fedora       1.5 MiB
 systemd-libs                      x86_64 257.4-3.fc43               fedora       2.2 MiB
 systemd-standalone-sysusers       x86_64 257.4-3.fc43               fedora     273.3 KiB
 tree-sitter-srpm-macros           noarch 0.2.0-1.fc43               fedora       6.9 KiB
 util-linux-core                   x86_64 2.40.4-7.fc43              fedora       1.4 MiB
 xxhash-libs                       x86_64 0.8.3-2.fc42               fedora      90.2 KiB
 xz-libs                           x86_64 1:5.6.3-3.fc42             fedora     218.3 KiB
 zig-srpm-macros                   noarch 1-4.fc42                   fedora       1.1 KiB
 zip                               x86_64 3.0-43.fc42                fedora     698.5 KiB
 zlib-ng-compat                    x86_64 2.2.4-2.fc43               fedora     137.6 KiB
 zstd                              x86_64 1.5.7-1.fc43               fedora       1.7 MiB
Installing groups:
 Buildsystem building group                                                              

Transaction Summary:
 Installing:       148 packages

Total size of inbound packages is 53 MiB. Need to download 0 B.
After this operation, 178 MiB extra will be used (install 178 MiB, remove 0 B).
[  1/148] tar-2:1.35-5.fc42.x86_64      100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[  2/148] bzip2-0:1.0.8-20.fc42.x86_64  100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[  3/148] redhat-rpm-config-0:342-2.fc4 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[  4/148] rpm-build-0:4.20.1-1.fc43.x86 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[  5/148] unzip-0:6.0-66.fc42.x86_64    100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[  6/148] cpio-0:2.15-2.fc41.x86_64     100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[  7/148] which-0:2.23-1.fc42.x86_64    100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[  8/148] bash-0:5.2.37-3.fc43.x86_64   100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[  9/148] coreutils-0:9.6-2.fc43.x86_64 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 10/148] grep-0:3.11-10.fc42.x86_64    100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 11/148] patch-0:2.7.6-26.fc42.x86_64  100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 12/148] sed-0:4.9-4.fc42.x86_64       100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 13/148] shadow-utils-2:4.17.4-1.fc43. 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 14/148] diffutils-0:3.10-9.fc42.x86_6 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 15/148] fedora-release-common-0:43-0. 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 16/148] findutils-1:4.10.0-5.fc42.x86 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 17/148] glibc-minimal-langpack-0:2.41 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 18/148] gzip-0:1.13-3.fc42.x86_64     100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 19/148] info-0:7.2-3.fc42.x86_64      100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 20/148] xz-1:5.6.3-3.fc42.x86_64      100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 21/148] util-linux-0:2.40.4-7.fc43.x8 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 22/148] gawk-0:5.3.1-1.fc42.x86_64    100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 23/148] glibc-0:2.41.9000-2.fc43.x86_ 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 24/148] libacl-0:2.3.2-3.fc42.x86_64  100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 25/148] libselinux-0:3.8-1.fc42.x86_6 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 26/148] bzip2-libs-0:1.0.8-20.fc42.x8 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 27/148] ansible-srpm-macros-0:1-17.1. 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 28/148] build-reproducibility-srpm-ma 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 29/148] dwz-0:0.15-9.fc42.x86_64      100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 30/148] efi-srpm-macros-0:6-3.fc43.no 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 31/148] file-0:5.46-1.fc42.x86_64     100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 32/148] filesystem-srpm-macros-0:3.18 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 33/148] fonts-srpm-macros-1:2.0.5-21. 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 34/148] forge-srpm-macros-0:0.4.0-2.f 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 35/148] fpc-srpm-macros-0:1.3-14.fc42 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 36/148] ghc-srpm-macros-0:1.9.2-2.fc4 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 37/148] gnat-srpm-macros-0:6-7.fc42.n 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 38/148] go-srpm-macros-0:3.6.0-6.fc42 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 39/148] kernel-srpm-macros-0:1.0-25.f 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 40/148] lua-srpm-macros-0:1-15.fc42.n 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 41/148] ocaml-srpm-macros-0:10-4.fc42 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 42/148] openblas-srpm-macros-0:2-19.f 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 43/148] package-notes-srpm-macros-0:0 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 44/148] perl-srpm-macros-0:1-57.fc42. 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 45/148] pyproject-srpm-macros-0:1.18. 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 46/148] python-srpm-macros-0:3.13-4.f 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 47/148] qt5-srpm-macros-0:5.15.16-1.f 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 48/148] qt6-srpm-macros-0:6.8.2-2.fc4 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 49/148] rpm-0:4.20.1-1.fc43.x86_64    100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 50/148] rust-srpm-macros-0:26.3-4.fc4 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 51/148] tree-sitter-srpm-macros-0:0.2 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 52/148] zig-srpm-macros-0:1-4.fc42.no 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 53/148] zip-0:3.0-43.fc42.x86_64      100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 54/148] debugedit-0:5.1-5.fc43.x86_64 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 55/148] elfutils-0:0.192-8.fc42.x86_6 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 56/148] elfutils-libelf-0:0.192-8.fc4 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 57/148] libarchive-0:3.7.7-4.fc43.x86 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 58/148] popt-0:1.19-8.fc42.x86_64     100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 59/148] readline-0:8.2-13.fc43.x86_64 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 60/148] rpm-build-libs-0:4.20.1-1.fc4 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 61/148] rpm-libs-0:4.20.1-1.fc43.x86_ 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 62/148] zstd-0:1.5.7-1.fc43.x86_64    100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 63/148] filesystem-0:3.18-38.fc43.x86 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 64/148] ncurses-libs-0:6.5-5.20250125 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 65/148] coreutils-common-0:9.6-2.fc43 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 66/148] gmp-1:6.3.0-3.fc43.x86_64     100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 67/148] libattr-0:2.5.2-5.fc42.x86_64 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 68/148] libcap-0:2.73-2.fc42.x86_64   100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 69/148] openssl-libs-1:3.5.0-1.fc43.x 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 70/148] systemd-libs-0:257.4-3.fc43.x 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 71/148] pcre2-0:10.45-1.fc43.x86_64   100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 72/148] ed-0:1.21-2.fc42.x86_64       100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 73/148] audit-libs-0:4.0.3-2.fc42.x86 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 74/148] libsemanage-0:3.8-1.fc42.x86_ 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 75/148] libxcrypt-0:4.4.38-6.fc43.x86 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 76/148] pam-libs-0:1.7.0-4.fc42.x86_6 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 77/148] setup-0:2.15.0-23.fc43.noarch 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 78/148] fedora-repos-0:43-0.1.noarch  100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 79/148] glibc-common-0:2.41.9000-2.fc 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 80/148] xz-libs-1:5.6.3-3.fc42.x86_64 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 81/148] libblkid-0:2.40.4-7.fc43.x86_ 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 82/148] libcap-ng-0:0.8.5-4.fc42.x86_ 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 83/148] libfdisk-0:2.40.4-7.fc43.x86_ 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 84/148] libmount-0:2.40.4-7.fc43.x86_ 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 85/148] libsmartcols-0:2.40.4-7.fc43. 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 86/148] libuuid-0:2.40.4-7.fc43.x86_6 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 87/148] util-linux-core-0:2.40.4-7.fc 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 88/148] zlib-ng-compat-0:2.2.4-2.fc43 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 89/148] mpfr-0:4.2.2-1.fc43.x86_64    100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 90/148] glibc-gconv-extra-0:2.41.9000 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 91/148] libgcc-0:15.0.1-0.10.fc43.x86 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 92/148] libsepol-0:3.8-1.fc42.x86_64  100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 93/148] add-determinism-0:0.6.0-1.fc4 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 94/148] file-libs-0:5.46-1.fc42.x86_6 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 95/148] curl-0:8.13.0~rc2-1.fc43.x86_ 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 96/148] elfutils-libs-0:0.192-8.fc42. 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 97/148] elfutils-debuginfod-client-0: 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 98/148] libstdc++-0:15.0.1-0.10.fc43. 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 99/148] libzstd-0:1.5.7-1.fc43.x86_64 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[100/148] libxml2-0:2.12.10-1.fc43.x86_ 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[101/148] lz4-libs-0:1.10.0-2.fc42.x86_ 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[102/148] libgomp-0:15.0.1-0.10.fc43.x8 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[103/148] lua-libs-0:5.4.7-3.fc43.x86_6 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[104/148] rpm-sequoia-0:1.7.0-5.fc43.x8 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[105/148] sqlite-libs-0:3.49.0-1.fc43.x 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[106/148] ncurses-base-0:6.5-5.20250125 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[107/148] ca-certificates-0:2024.2.69_v 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[108/148] crypto-policies-0:20250305-1. 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[109/148] pcre2-syntax-0:10.45-1.fc43.n 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[110/148] fedora-gpg-keys-0:43-0.1.noar 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[111/148] fedora-repos-rawhide-0:43-0.1 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[112/148] elfutils-default-yama-scope-0 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[113/148] json-c-0:0.18-2.fc42.x86_64   100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[114/148] libeconf-0:0.7.6-1.fc43.x86_6 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[115/148] binutils-0:2.44-3.fc43.x86_64 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[116/148] alternatives-0:1.32-1.fc43.x8 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[117/148] jansson-0:2.14-2.fc42.x86_64  100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[118/148] pkgconf-pkg-config-0:2.3.0-2. 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[119/148] pkgconf-0:2.3.0-2.fc42.x86_64 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[120/148] pkgconf-m4-0:2.3.0-2.fc42.noa 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[121/148] libpkgconf-0:2.3.0-2.fc42.x86 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[122/148] libffi-0:3.4.7-2.fc43.x86_64  100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[123/148] p11-kit-0:0.25.5-5.fc42.x86_6 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[124/148] libtasn1-0:4.20.0-1.fc43.x86_ 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[125/148] p11-kit-trust-0:0.25.5-5.fc42 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[126/148] fedora-release-0:43-0.7.noarc 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[127/148] systemd-standalone-sysusers-0 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[128/148] gdb-minimal-0:16.2-1.fc43.x86 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[129/148] xxhash-libs-0:0.8.3-2.fc42.x8 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[130/148] fedora-release-identity-basic 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[131/148] libcurl-0:8.13.0~rc2-1.fc43.x 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[132/148] krb5-libs-0:1.21.3-5.fc42.x86 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[133/148] libidn2-0:2.3.8-1.fc43.x86_64 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[134/148] libnghttp2-0:1.65.0-1.fc43.x8 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[135/148] libpsl-0:0.21.5-5.fc42.x86_64 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[136/148] libssh-0:0.11.1-4.fc42.x86_64 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[137/148] openldap-0:2.6.9-3.fc42.x86_6 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[138/148] keyutils-libs-0:1.6.3-5.fc42. 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[139/148] libcom_err-0:1.47.2-3.fc42.x8 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[140/148] libverto-0:0.3.2-10.fc42.x86_ 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[141/148] libunistring-0:1.1-9.fc42.x86 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[142/148] publicsuffix-list-dafsa-0:202 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[143/148] libssh-config-0:0.11.1-4.fc42 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[144/148] cyrus-sasl-lib-0:2.1.28-30.fc 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[145/148] libevent-0:2.1.12-15.fc42.x86 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[146/148] libtool-ltdl-0:2.5.4-4.fc42.x 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[147/148] gdbm-libs-1:1.23-9.fc42.x86_6 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[148/148] libbrotli-0:1.1.0-6.fc43.x86_ 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
--------------------------------------------------------------------------------
[148/148] Total                         100% |   0.0   B/s |   0.0   B |  00m00s
Running transaction
Importing OpenPGP key 0x31645531:
 UserID     : "Fedora (43) <fedora-43-primary@fedoraproject.org>"
 Fingerprint: C6E7F081CF80E13146676E88829B606631645531
 From       : file:///usr/share/distribution-gpg-keys/fedora/RPM-GPG-KEY-fedora-43-primary
The key was successfully imported.
Importing OpenPGP key 0x105EF944:
 UserID     : "Fedora (42) <fedora-42-primary@fedoraproject.org>"
 Fingerprint: B0F4950458F69E1150C6C5EDC8AC4916105EF944
 From       : file:///usr/share/distribution-gpg-keys/fedora/RPM-GPG-KEY-fedora-42-primary
The key was successfully imported.
Importing OpenPGP key 0x6D9F90A6:
 UserID     : "Fedora (44) <fedora-44-primary@fedoraproject.org>"
 Fingerprint: 36F612DCF27F7D1A48A835E4DBFCF71C6D9F90A6
 From       : file:///usr/share/distribution-gpg-keys/fedora/RPM-GPG-KEY-fedora-44-primary
The key was successfully imported.
[  1/150] Verify package files          100% | 751.0   B/s | 148.0   B |  00m00s
>>> Running pre-transaction scriptlet: filesystem-0:3.18-38.fc43.x86_64         
>>> Finished pre-transaction scriptlet: filesystem-0:3.18-38.fc43.x86_64        
>>> [RPM] /var/lib/mock/fedora-rawhide-x86_64-1742774257.650632/root/var/cache/d
[  2/150] Prepare transaction           100% |   1.9 KiB/s | 148.0   B |  00m00s
[  3/150] Installing libgcc-0:15.0.1-0. 100% | 131.0 MiB/s | 268.3 KiB |  00m00s
[  4/150] Installing libssh-config-0:0. 100% |   0.0   B/s | 816.0   B |  00m00s
[  5/150] Installing publicsuffix-list- 100% |  67.6 MiB/s |  69.2 KiB |  00m00s
[  6/150] Installing fedora-release-ide 100% | 953.1 KiB/s | 976.0   B |  00m00s
[  7/150] Installing fedora-gpg-keys-0: 100% |  19.0 MiB/s | 174.8 KiB |  00m00s
[  8/150] Installing fedora-repos-rawhi 100% |   0.0   B/s |   2.4 KiB |  00m00s
[  9/150] Installing fedora-repos-0:43- 100% |   0.0   B/s |   5.7 KiB |  00m00s
[ 10/150] Installing fedora-release-com 100% |  12.0 MiB/s |  24.6 KiB |  00m00s
[ 11/150] Installing fedora-release-0:4 100% |   9.3 KiB/s | 124.0   B |  00m00s
>>> Running unknown scriptlet: setup-0:2.15.0-23.fc43.noarch                    
>>> Finished unknown scriptlet: setup-0:2.15.0-23.fc43.noarch                   
>>> Scriptlet output:                                                           
>>> Creating group 'adm' with GID 4.                                            
>>> Creating group 'audio' with GID 63.                                         
>>> Creating group 'cdrom' with GID 11.                                         
>>> Creating group 'clock' with GID 103.                                        
>>> Creating group 'dialout' with GID 18.                                       
>>> Creating group 'disk' with GID 6.                                           
>>> Creating group 'floppy' with GID 19.                                        
>>> Creating group 'ftp' with GID 50.                                           
>>> Creating group 'games' with GID 20.                                         
>>> Creating group 'input' with GID 104.                                        
>>> Creating group 'kmem' with GID 9.                                           
>>> Creating group 'kvm' with GID 36.                                           
>>> Creating group 'lock' with GID 54.                                          
>>> Creating group 'lp' with GID 7.                                             
>>> Creating group 'mail' with GID 12.                                          
>>> Creating group 'man' with GID 15.                                           
>>> Creating group 'mem' with GID 8.                                            
>>> Creating group 'nobody' with GID 65534.                                     
>>> Creating group 'render' with GID 105.                                       
>>> Creating group 'root' with GID 0.                                           
>>> Creating group 'sgx' with GID 106.                                          
>>> Creating group 'sys' with GID 3.                                            
>>> Creating group 'tape' with GID 33.                                          
>>> Creating group 'tty' with GID 5.                                            
>>> Creating group 'users' with GID 100.                                        
>>> Creating group 'utmp' with GID 22.                                          
>>> Creating group 'video' with GID 39.                                         
>>> Creating group 'wheel' with GID 10.                                         
>>> Creating user 'adm' (adm) with UID 3 and GID 4.                             
>>> Creating group 'bin' with GID 1.                                            
>>> Creating user 'bin' (bin) with UID 1 and GID 1.                             
>>> Creating group 'daemon' with GID 2.                                         
>>> Creating user 'daemon' (daemon) with UID 2 and GID 2.                       
>>> Creating user 'ftp' (FTP User) with UID 14 and GID 50.                      
>>> Creating user 'games' (games) with UID 12 and GID 100.                      
>>> Creating user 'halt' (halt) with UID 7 and GID 0.                           
>>> Creating user 'lp' (lp) with UID 4 and GID 7.                               
>>> Creating user 'mail' (mail) with UID 8 and GID 12.                          
>>> Creating user 'nobody' (Kernel Overflow User) with UID 65534 and GID 65534. 
>>> Creating user 'operator' (operator) with UID 11 and GID 0.                  
>>> Creating user 'root' (Super User) with UID 0 and GID 0.                     
>>> Creating user 'shutdown' (shutdown) with UID 6 and GID 0.                   
>>> Creating user 'sync' (sync) with UID 5 and GID 0.                           
>>>                                                                             
[ 12/150] Installing setup-0:2.15.0-23. 100% |  39.6 MiB/s | 730.3 KiB |  00m00s
>>> [RPM] /etc/hosts created as /etc/hosts.rpmnew                               
[ 13/150] Installing filesystem-0:3.18- 100% |   1.4 MiB/s | 212.4 KiB |  00m00s
[ 14/150] Installing pkgconf-m4-0:2.3.0 100% |  14.5 MiB/s |  14.8 KiB |  00m00s
[ 15/150] Installing pcre2-syntax-0:10. 100% | 135.0 MiB/s | 276.4 KiB |  00m00s
[ 16/150] Installing ncurses-base-0:6.5 100% |  38.2 MiB/s | 352.2 KiB |  00m00s
[ 17/150] Installing glibc-minimal-lang 100% |   0.0   B/s | 124.0   B |  00m00s
[ 18/150] Installing ncurses-libs-0:6.5 100% | 132.9 MiB/s | 952.8 KiB |  00m00s
[ 19/150] Installing glibc-0:2.41.9000- 100% | 151.5 MiB/s |   6.7 MiB |  00m00s
[ 20/150] Installing bash-0:5.2.37-3.fc 100% | 199.5 MiB/s |   8.2 MiB |  00m00s
[ 21/150] Installing glibc-common-0:2.4 100% |  51.0 MiB/s |   1.0 MiB |  00m00s
[ 22/150] Installing glibc-gconv-extra- 100% | 140.6 MiB/s |   7.3 MiB |  00m00s
[ 23/150] Installing zlib-ng-compat-0:2 100% | 135.2 MiB/s | 138.4 KiB |  00m00s
[ 24/150] Installing bzip2-libs-0:1.0.8 100% |  83.7 MiB/s |  85.7 KiB |  00m00s
[ 25/150] Installing xz-libs-1:5.6.3-3. 100% | 214.3 MiB/s | 219.4 KiB |  00m00s
[ 26/150] Installing libuuid-0:2.40.4-7 100% |  37.5 MiB/s |  38.4 KiB |  00m00s
[ 27/150] Installing libblkid-0:2.40.4- 100% | 128.6 MiB/s | 263.4 KiB |  00m00s
[ 28/150] Installing popt-0:1.19-8.fc42 100% |  27.2 MiB/s | 139.4 KiB |  00m00s
[ 29/150] Installing readline-0:8.2-13. 100% | 158.6 MiB/s | 487.1 KiB |  00m00s
[ 30/150] Installing gmp-1:6.3.0-3.fc43 100% | 267.4 MiB/s | 821.5 KiB |  00m00s
[ 31/150] Installing libxcrypt-0:4.4.38 100% | 140.2 MiB/s | 287.2 KiB |  00m00s
[ 32/150] Installing libstdc++-0:15.0.1 100% | 281.2 MiB/s |   2.8 MiB |  00m00s
[ 33/150] Installing libzstd-0:1.5.7-1. 100% | 263.4 MiB/s | 809.1 KiB |  00m00s
[ 34/150] Installing elfutils-libelf-0: 100% | 292.5 MiB/s |   1.2 MiB |  00m00s
[ 35/150] Installing libattr-0:2.5.2-5. 100% |  27.4 MiB/s |  28.1 KiB |  00m00s
[ 36/150] Installing libacl-0:2.3.2-3.f 100% |  38.2 MiB/s |  39.2 KiB |  00m00s
[ 37/150] Installing dwz-0:0.15-9.fc42. 100% |  20.4 MiB/s | 292.4 KiB |  00m00s
[ 38/150] Installing mpfr-0:4.2.2-1.fc4 100% | 202.7 MiB/s | 830.4 KiB |  00m00s
[ 39/150] Installing gawk-0:5.3.1-1.fc4 100% |  77.0 MiB/s |   1.7 MiB |  00m00s
[ 40/150] Installing unzip-0:6.0-66.fc4 100% |  27.5 MiB/s | 393.8 KiB |  00m00s
[ 41/150] Installing file-libs-0:5.46-1 100% | 515.5 MiB/s |  11.9 MiB |  00m00s
[ 42/150] Installing file-0:5.46-1.fc42 100% |   4.0 MiB/s | 101.7 KiB |  00m00s
[ 43/150] Installing crypto-policies-0: 100% |  15.8 MiB/s | 161.4 KiB |  00m00s
[ 44/150] Installing pcre2-0:10.45-1.fc 100% | 227.6 MiB/s | 699.1 KiB |  00m00s
[ 45/150] Installing grep-0:3.11-10.fc4 100% |  45.6 MiB/s |   1.0 MiB |  00m00s
[ 46/150] Installing xz-1:5.6.3-3.fc42. 100% |  58.5 MiB/s |   1.2 MiB |  00m00s
[ 47/150] Installing libcap-ng-0:0.8.5- 100% |  73.1 MiB/s |  74.8 KiB |  00m00s
[ 48/150] Installing audit-libs-0:4.0.3 100% | 172.6 MiB/s | 353.4 KiB |  00m00s
[ 49/150] Installing libsmartcols-0:2.4 100% | 177.3 MiB/s | 181.5 KiB |  00m00s
[ 50/150] Installing libsepol-0:3.8-1.f 100% | 269.2 MiB/s | 827.0 KiB |  00m00s
[ 51/150] Installing libselinux-0:3.8-1 100% |  94.9 MiB/s | 194.3 KiB |  00m00s
[ 52/150] Installing sed-0:4.9-4.fc42.x 100% |  44.5 MiB/s | 865.5 KiB |  00m00s
[ 53/150] Installing findutils-1:4.10.0 100% |  89.2 MiB/s |   1.9 MiB |  00m00s
[ 54/150] Installing libmount-0:2.40.4- 100% | 174.5 MiB/s | 357.4 KiB |  00m00s
[ 55/150] Installing lz4-libs-0:1.10.0- 100% | 154.7 MiB/s | 158.5 KiB |  00m00s
[ 56/150] Installing lua-libs-0:5.4.7-3 100% | 135.8 MiB/s | 278.1 KiB |  00m00s
[ 57/150] Installing libeconf-0:0.7.6-1 100% |  64.7 MiB/s |  66.2 KiB |  00m00s
[ 58/150] Installing pam-libs-0:1.7.0-4 100% |  63.1 MiB/s | 129.1 KiB |  00m00s
[ 59/150] Installing libcap-0:2.73-2.fc 100% |  13.8 MiB/s | 212.1 KiB |  00m00s
[ 60/150] Installing systemd-libs-0:257 100% | 278.0 MiB/s |   2.2 MiB |  00m00s
[ 61/150] Installing alternatives-0:1.3 100% |   4.8 MiB/s |  63.8 KiB |  00m00s
[ 62/150] Installing libffi-0:3.4.7-2.f 100% |  82.0 MiB/s |  84.0 KiB |  00m00s
[ 63/150] Installing libtasn1-0:4.20.0- 100% |  87.0 MiB/s | 178.1 KiB |  00m00s
[ 64/150] Installing p11-kit-0:0.25.5-5 100% |  84.0 MiB/s |   2.2 MiB |  00m00s
[ 65/150] Installing libunistring-0:1.1 100% | 287.8 MiB/s |   1.7 MiB |  00m00s
[ 66/150] Installing libidn2-0:2.3.8-1. 100% | 109.1 MiB/s | 558.7 KiB |  00m00s
[ 67/150] Installing libpsl-0:0.21.5-5. 100% |  75.7 MiB/s |  77.5 KiB |  00m00s
[ 68/150] Installing p11-kit-trust-0:0. 100% |  14.4 MiB/s | 397.2 KiB |  00m00s
[ 69/150] Installing util-linux-core-0: 100% |  59.4 MiB/s |   1.4 MiB |  00m00s
[ 70/150] Installing systemd-standalone 100% |  20.6 MiB/s | 273.8 KiB |  00m00s
[ 71/150] Installing zstd-0:1.5.7-1.fc4 100% |  95.0 MiB/s |   1.7 MiB |  00m00s
[ 72/150] Installing tar-2:1.35-5.fc42. 100% | 118.5 MiB/s |   3.0 MiB |  00m00s
[ 73/150] Installing libsemanage-0:3.8- 100% | 101.0 MiB/s | 310.2 KiB |  00m00s
[ 74/150] Installing shadow-utils-2:4.1 100% |  88.1 MiB/s |   4.1 MiB |  00m00s
[ 75/150] Installing zip-0:3.0-43.fc42. 100% |  42.9 MiB/s | 702.4 KiB |  00m00s
[ 76/150] Installing libfdisk-0:2.40.4- 100% | 182.4 MiB/s | 373.5 KiB |  00m00s
[ 77/150] Installing libxml2-0:2.12.10- 100% |  89.3 MiB/s |   1.7 MiB |  00m00s
[ 78/150] Installing bzip2-0:1.0.8-20.f 100% |   7.8 MiB/s | 103.8 KiB |  00m00s
[ 79/150] Installing add-determinism-0: 100% | 123.3 MiB/s |   2.5 MiB |  00m00s
[ 80/150] Installing build-reproducibil 100% |   0.0   B/s |   1.0 KiB |  00m00s
[ 81/150] Installing filesystem-srpm-ma 100% |  38.0 MiB/s |  38.9 KiB |  00m00s
[ 82/150] Installing ed-0:1.21-2.fc42.x 100% |  11.2 MiB/s | 148.8 KiB |  00m00s
[ 83/150] Installing patch-0:2.7.6-26.f 100% |  19.5 MiB/s | 260.2 KiB |  00m00s
[ 84/150] Installing elfutils-default-y 100% | 185.7 KiB/s |   2.0 KiB |  00m00s
[ 85/150] Installing elfutils-libs-0:0. 100% | 165.2 MiB/s | 676.7 KiB |  00m00s
[ 86/150] Installing cpio-0:2.15-2.fc41 100% |  52.4 MiB/s |   1.1 MiB |  00m00s
[ 87/150] Installing diffutils-0:3.10-9 100% |  75.7 MiB/s |   1.6 MiB |  00m00s
[ 88/150] Installing libgomp-0:15.0.1-0 100% | 262.4 MiB/s | 537.3 KiB |  00m00s
[ 89/150] Installing sqlite-libs-0:3.49 100% | 254.0 MiB/s |   1.5 MiB |  00m00s
[ 90/150] Installing json-c-0:0.18-2.fc 100% |  85.9 MiB/s |  88.0 KiB |  00m00s
[ 91/150] Installing jansson-0:2.14-2.f 100% |  92.2 MiB/s |  94.4 KiB |  00m00s
[ 92/150] Installing libpkgconf-0:2.3.0 100% |  77.4 MiB/s |  79.2 KiB |  00m00s
[ 93/150] Installing pkgconf-0:2.3.0-2. 100% |   6.8 MiB/s |  91.0 KiB |  00m00s
[ 94/150] Installing pkgconf-pkg-config 100% | 147.8 KiB/s |   1.8 KiB |  00m00s
[ 95/150] Installing xxhash-libs-0:0.8. 100% |  89.4 MiB/s |  91.6 KiB |  00m00s
[ 96/150] Installing libnghttp2-0:1.65. 100% | 159.5 MiB/s | 163.3 KiB |  00m00s
[ 97/150] Installing keyutils-libs-0:1. 100% |  58.3 MiB/s |  59.7 KiB |  00m00s
[ 98/150] Installing libcom_err-0:1.47. 100% |  66.6 MiB/s |  68.2 KiB |  00m00s
[ 99/150] Installing libverto-0:0.3.2-1 100% |  26.6 MiB/s |  27.2 KiB |  00m00s
[100/150] Installing libtool-ltdl-0:2.5 100% |  69.6 MiB/s |  71.2 KiB |  00m00s
[101/150] Installing gdbm-libs-1:1.23-9 100% | 128.5 MiB/s | 131.6 KiB |  00m00s
[102/150] Installing cyrus-sasl-lib-0:2 100% | 109.7 MiB/s |   2.3 MiB |  00m00s
[103/150] Installing libbrotli-0:1.1.0- 100% | 162.4 MiB/s | 831.6 KiB |  00m00s
[104/150] Installing coreutils-common-0 100% | 247.9 MiB/s |  11.2 MiB |  00m00s
[105/150] Installing openssl-libs-1:3.5 100% | 317.3 MiB/s |   8.9 MiB |  00m00s
[106/150] Installing coreutils-0:9.6-2. 100% | 109.1 MiB/s |   5.5 MiB |  00m00s
[107/150] Installing ca-certificates-0: 100% |   1.2 MiB/s |   2.4 MiB |  00m02s
[108/150] Installing libarchive-0:3.7.7 100% | 182.2 MiB/s | 932.6 KiB |  00m00s
[109/150] Installing krb5-libs-0:1.21.3 100% | 191.6 MiB/s |   2.3 MiB |  00m00s
[110/150] Installing libssh-0:0.11.1-4. 100% | 184.7 MiB/s | 567.5 KiB |  00m00s
[111/150] Installing gzip-0:1.13-3.fc42 100% |  24.3 MiB/s | 398.4 KiB |  00m00s
[112/150] Installing rpm-sequoia-0:1.7. 100% | 268.3 MiB/s |   2.4 MiB |  00m00s
[113/150] Installing rpm-libs-0:4.20.1- 100% | 176.6 MiB/s | 723.4 KiB |  00m00s
[114/150] Installing rpm-build-libs-0:4 100% | 202.6 MiB/s | 207.4 KiB |  00m00s
[115/150] Installing libevent-0:2.1.12- 100% | 221.4 MiB/s | 906.9 KiB |  00m00s
[116/150] Installing openldap-0:2.6.9-3 100% | 160.9 MiB/s | 658.9 KiB |  00m00s
[117/150] Installing libcurl-0:8.13.0~r 100% | 212.8 MiB/s | 871.5 KiB |  00m00s
[118/150] Installing elfutils-debuginfo 100% |   6.0 MiB/s |  86.2 KiB |  00m00s
[119/150] Installing elfutils-0:0.192-8 100% | 116.8 MiB/s |   2.7 MiB |  00m00s
[120/150] Installing binutils-0:2.44-3. 100% | 239.8 MiB/s |  25.9 MiB |  00m00s
[121/150] Installing gdb-minimal-0:16.2 100% | 241.8 MiB/s |  13.3 MiB |  00m00s
[122/150] Installing debugedit-0:5.1-5. 100% |  13.6 MiB/s | 195.4 KiB |  00m00s
[123/150] Installing curl-0:8.13.0~rc2- 100% |  15.0 MiB/s | 459.5 KiB |  00m00s
[124/150] Installing rpm-0:4.20.1-1.fc4 100% |  60.9 MiB/s |   2.5 MiB |  00m00s
[125/150] Installing efi-srpm-macros-0: 100% |  40.2 MiB/s |  41.1 KiB |  00m00s
[126/150] Installing lua-srpm-macros-0: 100% |   1.9 MiB/s |   1.9 KiB |  00m00s
[127/150] Installing tree-sitter-srpm-m 100% |   7.8 MiB/s |   7.9 KiB |  00m00s
[128/150] Installing zig-srpm-macros-0: 100% |   1.6 MiB/s |   1.7 KiB |  00m00s
[129/150] Installing rust-srpm-macros-0 100% |   0.0   B/s |   5.6 KiB |  00m00s
[130/150] Installing qt6-srpm-macros-0: 100% |   0.0   B/s | 740.0   B |  00m00s
[131/150] Installing qt5-srpm-macros-0: 100% |   0.0   B/s | 776.0   B |  00m00s
[132/150] Installing perl-srpm-macros-0 100% |   0.0   B/s |   1.1 KiB |  00m00s
[133/150] Installing package-notes-srpm 100% |   0.0   B/s |   2.0 KiB |  00m00s
[134/150] Installing openblas-srpm-macr 100% |   0.0   B/s | 392.0   B |  00m00s
[135/150] Installing ocaml-srpm-macros- 100% |   0.0   B/s |   2.2 KiB |  00m00s
[136/150] Installing kernel-srpm-macros 100% |   0.0   B/s |   2.3 KiB |  00m00s
[137/150] Installing gnat-srpm-macros-0 100% |   0.0   B/s |   1.3 KiB |  00m00s
[138/150] Installing ghc-srpm-macros-0: 100% |   0.0   B/s |   1.0 KiB |  00m00s
[139/150] Installing fpc-srpm-macros-0: 100% |   0.0   B/s | 420.0   B |  00m00s
[140/150] Installing ansible-srpm-macro 100% |  35.4 MiB/s |  36.2 KiB |  00m00s
[141/150] Installing fonts-srpm-macros- 100% |  55.7 MiB/s |  57.0 KiB |  00m00s
[142/150] Installing forge-srpm-macros- 100% |  39.3 MiB/s |  40.3 KiB |  00m00s
[143/150] Installing go-srpm-macros-0:3 100% |  60.5 MiB/s |  62.0 KiB |  00m00s
[144/150] Installing python-srpm-macros 100% |  50.9 MiB/s |  52.2 KiB |  00m00s
[145/150] Installing redhat-rpm-config- 100% |  47.2 MiB/s | 193.5 KiB |  00m00s
[146/150] Installing rpm-build-0:4.20.1 100% |  10.2 MiB/s | 177.4 KiB |  00m00s
[147/150] Installing pyproject-srpm-mac 100% |   2.4 MiB/s |   2.5 KiB |  00m00s
[148/150] Installing which-0:2.23-1.fc4 100% |   6.0 MiB/s |  85.6 KiB |  00m00s
[149/150] Installing util-linux-0:2.40. 100% |  65.3 MiB/s |   3.5 MiB |  00m00s
[150/150] Installing info-0:7.2-3.fc42. 100% | 134.0 KiB/s | 358.3 KiB |  00m03s
Public key "file:///usr/share/distribution-gpg-keys/fedora/RPM-GPG-KEY-fedora-43-primary" is already present, not importing.
Warning: skipped OpenPGP checks for 2 packages from repository: copr_base
Complete!
Finish: installing minimal buildroot with dnf5
Start: creating root cache
Finish: creating root cache
Finish: chroot init
INFO: Installed packages:
INFO: add-determinism-0.6.0-1.fc43.x86_64
alternatives-1.32-1.fc43.x86_64
ansible-srpm-macros-1-17.1.fc42.noarch
audit-libs-4.0.3-2.fc42.x86_64
bash-5.2.37-3.fc43.x86_64
binutils-2.44-3.fc43.x86_64
build-reproducibility-srpm-macros-0.6.0-1.fc43.noarch
bzip2-1.0.8-20.fc42.x86_64
bzip2-libs-1.0.8-20.fc42.x86_64
ca-certificates-2024.2.69_v8.0.401-5.fc42.noarch
coreutils-9.6-2.fc43.x86_64
coreutils-common-9.6-2.fc43.x86_64
cpio-2.15-2.fc41.x86_64
crypto-policies-20250305-1.gita35b0fa.fc43.noarch
curl-8.13.0~rc2-1.fc43.x86_64
cyrus-sasl-lib-2.1.28-30.fc42.x86_64
debugedit-5.1-5.fc43.x86_64
diffutils-3.10-9.fc42.x86_64
dwz-0.15-9.fc42.x86_64
ed-1.21-2.fc42.x86_64
efi-srpm-macros-6-3.fc43.noarch
elfutils-0.192-8.fc42.x86_64
elfutils-debuginfod-client-0.192-8.fc42.x86_64
elfutils-default-yama-scope-0.192-8.fc42.noarch
elfutils-libelf-0.192-8.fc42.x86_64
elfutils-libs-0.192-8.fc42.x86_64
fedora-gpg-keys-43-0.1.noarch
fedora-release-43-0.7.noarch
fedora-release-common-43-0.7.noarch
fedora-release-identity-basic-43-0.7.noarch
fedora-repos-43-0.1.noarch
fedora-repos-rawhide-43-0.1.noarch
file-5.46-1.fc42.x86_64
file-libs-5.46-1.fc42.x86_64
filesystem-3.18-38.fc43.x86_64
filesystem-srpm-macros-3.18-38.fc43.noarch
findutils-4.10.0-5.fc42.x86_64
fonts-srpm-macros-2.0.5-21.fc42.noarch
forge-srpm-macros-0.4.0-2.fc42.noarch
fpc-srpm-macros-1.3-14.fc42.noarch
gawk-5.3.1-1.fc42.x86_64
gdb-minimal-16.2-1.fc43.x86_64
gdbm-libs-1.23-9.fc42.x86_64
ghc-srpm-macros-1.9.2-2.fc42.noarch
glibc-2.41.9000-2.fc43.x86_64
glibc-common-2.41.9000-2.fc43.x86_64
glibc-gconv-extra-2.41.9000-2.fc43.x86_64
glibc-minimal-langpack-2.41.9000-2.fc43.x86_64
gmp-6.3.0-3.fc43.x86_64
gnat-srpm-macros-6-7.fc42.noarch
go-srpm-macros-3.6.0-6.fc42.noarch
gpg-pubkey-105ef944-65ca83d1
gpg-pubkey-31645531-66b6dccf
gpg-pubkey-6d9f90a6-6786af3b
grep-3.11-10.fc42.x86_64
gzip-1.13-3.fc42.x86_64
info-7.2-3.fc42.x86_64
jansson-2.14-2.fc42.x86_64
json-c-0.18-2.fc42.x86_64
kernel-srpm-macros-1.0-25.fc42.noarch
keyutils-libs-1.6.3-5.fc42.x86_64
krb5-libs-1.21.3-5.fc42.x86_64
libacl-2.3.2-3.fc42.x86_64
libarchive-3.7.7-4.fc43.x86_64
libattr-2.5.2-5.fc42.x86_64
libblkid-2.40.4-7.fc43.x86_64
libbrotli-1.1.0-6.fc43.x86_64
libcap-2.73-2.fc42.x86_64
libcap-ng-0.8.5-4.fc42.x86_64
libcom_err-1.47.2-3.fc42.x86_64
libcurl-8.13.0~rc2-1.fc43.x86_64
libeconf-0.7.6-1.fc43.x86_64
libevent-2.1.12-15.fc42.x86_64
libfdisk-2.40.4-7.fc43.x86_64
libffi-3.4.7-2.fc43.x86_64
libgcc-15.0.1-0.10.fc43.x86_64
libgomp-15.0.1-0.10.fc43.x86_64
libidn2-2.3.8-1.fc43.x86_64
libmount-2.40.4-7.fc43.x86_64
libnghttp2-1.65.0-1.fc43.x86_64
libpkgconf-2.3.0-2.fc42.x86_64
libpsl-0.21.5-5.fc42.x86_64
libselinux-3.8-1.fc42.x86_64
libsemanage-3.8-1.fc42.x86_64
libsepol-3.8-1.fc42.x86_64
libsmartcols-2.40.4-7.fc43.x86_64
libssh-0.11.1-4.fc42.x86_64
libssh-config-0.11.1-4.fc42.noarch
libstdc++-15.0.1-0.10.fc43.x86_64
libtasn1-4.20.0-1.fc43.x86_64
libtool-ltdl-2.5.4-4.fc42.x86_64
libunistring-1.1-9.fc42.x86_64
libuuid-2.40.4-7.fc43.x86_64
libverto-0.3.2-10.fc42.x86_64
libxcrypt-4.4.38-6.fc43.x86_64
libxml2-2.12.10-1.fc43.x86_64
libzstd-1.5.7-1.fc43.x86_64
lua-libs-5.4.7-3.fc43.x86_64
lua-srpm-macros-1-15.fc42.noarch
lz4-libs-1.10.0-2.fc42.x86_64
mpfr-4.2.2-1.fc43.x86_64
ncurses-base-6.5-5.20250125.fc42.noarch
ncurses-libs-6.5-5.20250125.fc42.x86_64
ocaml-srpm-macros-10-4.fc42.noarch
openblas-srpm-macros-2-19.fc42.noarch
openldap-2.6.9-3.fc42.x86_64
openssl-libs-3.5.0-1.fc43.x86_64
p11-kit-0.25.5-5.fc42.x86_64
p11-kit-trust-0.25.5-5.fc42.x86_64
package-notes-srpm-macros-0.5-13.fc42.noarch
pam-libs-1.7.0-4.fc42.x86_64
patch-2.7.6-26.fc42.x86_64
pcre2-10.45-1.fc43.x86_64
pcre2-syntax-10.45-1.fc43.noarch
perl-srpm-macros-1-57.fc42.noarch
pkgconf-2.3.0-2.fc42.x86_64
pkgconf-m4-2.3.0-2.fc42.noarch
pkgconf-pkg-config-2.3.0-2.fc42.x86_64
popt-1.19-8.fc42.x86_64
publicsuffix-list-dafsa-20250116-1.fc42.noarch
pyproject-srpm-macros-1.18.1-1.fc43.noarch
python-srpm-macros-3.13-4.fc42.noarch
qt5-srpm-macros-5.15.16-1.fc43.noarch
qt6-srpm-macros-6.8.2-2.fc43.noarch
readline-8.2-13.fc43.x86_64
redhat-rpm-config-342-2.fc42.noarch
rpm-4.20.1-1.fc43.x86_64
rpm-build-4.20.1-1.fc43.x86_64
rpm-build-libs-4.20.1-1.fc43.x86_64
rpm-libs-4.20.1-1.fc43.x86_64
rpm-sequoia-1.7.0-5.fc43.x86_64
rust-srpm-macros-26.3-4.fc42.noarch
sed-4.9-4.fc42.x86_64
setup-2.15.0-23.fc43.noarch
shadow-utils-4.17.4-1.fc43.x86_64
sqlite-libs-3.49.0-1.fc43.x86_64
systemd-libs-257.4-3.fc43.x86_64
systemd-standalone-sysusers-257.4-3.fc43.x86_64
tar-1.35-5.fc42.x86_64
tree-sitter-srpm-macros-0.2.0-1.fc43.noarch
unzip-6.0-66.fc42.x86_64
util-linux-2.40.4-7.fc43.x86_64
util-linux-core-2.40.4-7.fc43.x86_64
which-2.23-1.fc42.x86_64
xxhash-libs-0.8.3-2.fc42.x86_64
xz-5.6.3-3.fc42.x86_64
xz-libs-5.6.3-3.fc42.x86_64
zig-srpm-macros-1-4.fc42.noarch
zip-3.0-43.fc42.x86_64
zlib-ng-compat-2.2.4-2.fc43.x86_64
zstd-1.5.7-1.fc43.x86_64
Start: buildsrpm
Start: rpmbuild -bs
Building target platforms: x86_64
Building for target x86_64
setting SOURCE_DATE_EPOCH=1741478400
Wrote: /builddir/build/SRPMS/llama-cpp-b4580-2.fc43.src.rpm
Finish: rpmbuild -bs
INFO: chroot_scan: 1 files copied to /var/lib/copr-rpmbuild/results/chroot_scan
INFO: /var/lib/mock/fedora-rawhide-x86_64-1742774257.650632/root/var/log/dnf5.log
INFO: chroot_scan: creating tarball /var/lib/copr-rpmbuild/results/chroot_scan.tar.gz
/bin/tar: Removing leading `/' from member names
Finish: buildsrpm
INFO: Done(/var/lib/copr-rpmbuild/workspace/workdir-s56ti4b2/llama-cpp/llama-cpp.spec) Config(child) 0 minutes 17 seconds
INFO: Results and/or logs in: /var/lib/copr-rpmbuild/results
INFO: Cleaning up build root ('cleanup_on_success=True')
Start: clean chroot
INFO: unmounting tmpfs.
Finish: clean chroot
INFO: Start(/var/lib/copr-rpmbuild/results/llama-cpp-b4580-2.fc43.src.rpm)  Config(fedora-rawhide-x86_64)
Start(bootstrap): chroot init
INFO: mounting tmpfs at /var/lib/mock/fedora-rawhide-x86_64-bootstrap-1742774257.650632/root.
INFO: reusing tmpfs at /var/lib/mock/fedora-rawhide-x86_64-bootstrap-1742774257.650632/root.
INFO: calling preinit hooks
INFO: enabled root cache
INFO: enabled package manager cache
Start(bootstrap): cleaning package manager metadata
Finish(bootstrap): cleaning package manager metadata
Finish(bootstrap): chroot init
Start: chroot init
INFO: mounting tmpfs at /var/lib/mock/fedora-rawhide-x86_64-1742774257.650632/root.
INFO: calling preinit hooks
INFO: enabled root cache
Start: unpacking root cache
Finish: unpacking root cache
INFO: enabled package manager cache
Start: cleaning package manager metadata
Finish: cleaning package manager metadata
INFO: enabled HW Info plugin
INFO: Buildroot is handled by package management downloaded with a bootstrap image:
  rpm-4.20.1-1.fc43.x86_64
  rpm-sequoia-1.7.0-5.fc43.x86_64
  dnf5-5.2.12.0-1.fc43.x86_64
  dnf5-plugins-5.2.12.0-1.fc43.x86_64
Finish: chroot init
Start: build phase for llama-cpp-b4580-2.fc43.src.rpm
Start: build setup for llama-cpp-b4580-2.fc43.src.rpm
Building target platforms: x86_64
Building for target x86_64
setting SOURCE_DATE_EPOCH=1741478400
Wrote: /builddir/build/SRPMS/llama-cpp-b4580-2.fc43.src.rpm
Updating and loading repositories:
 fedora                                 100% | 406.0 KiB/s |  14.2 KiB |  00m00s
 Copr repository                        100% |  26.9 KiB/s |   1.5 KiB |  00m00s
 Copr repository                        100% |  13.5 MiB/s |   1.6 MiB |  00m00s
Repositories loaded.
Package "curl-8.13.0~rc2-1.fc43.x86_64" is already installed.

Package                         Arch   Version                        Repository      Size
Installing:
 cmake                          x86_64 4.0.0~rc4-2.fc43               copr_base   34.4 MiB
 gcc-c++                        x86_64 15.0.1-0.10.fc43               fedora      41.1 MiB
 git                            x86_64 2.49.0-1.fc43                  fedora      85.3 KiB
 hipblas-devel                  x86_64 6.3.0-5.fc43                   copr_base    3.1 MiB
 hipcc-libomp-devel             x86_64 18-43.rocm6.3.2.fc43           fedora       0.0   B
 langpacks-en                   noarch 4.2-4.fc42                     fedora     400.0   B
 libcurl-devel                  x86_64 8.13.0~rc2-1.fc43              fedora       1.3 MiB
 openmpi                        x86_64 5.0.6-5.fc43                   fedora       7.0 MiB
 pthreadpool-devel              x86_64 0.0^git20230829.4fe0e1e-6.fc42 fedora      99.1 KiB
 rocblas-devel                  x86_64 6.3.0-7.fc43                   fedora       2.8 MiB
 rocm-comgr-devel               x86_64 18-43.rocm6.3.2.fc43           fedora     103.0 KiB
 rocm-hip-devel                 x86_64 6.3.2-3.fc43                   fedora       2.7 MiB
 rocm-rpm-macros                noarch 6.3.2-2.fc43                   fedora      18.9 KiB
 rocm-runtime-devel             x86_64 6.3.2-5.fc43                   fedora     565.6 KiB
 wget2-wget                     x86_64 2.2.0-3.fc43                   fedora      42.0   B
 xxd                            x86_64 2:9.1.1227-1.fc43              fedora      33.3 KiB
Installing dependencies:
 abattis-cantarell-vf-fonts     noarch 0.301-14.fc42                  fedora     192.7 KiB
 annobin-docs                   noarch 12.93-1.fc43                   fedora      98.9 KiB
 annobin-plugin-gcc             x86_64 12.93-1.fc43                   fedora     993.4 KiB
 brotli                         x86_64 1.1.0-6.fc43                   copr_base   31.6 KiB
 brotli-devel                   x86_64 1.1.0-6.fc43                   copr_base   65.6 KiB
 clang-resource-filesystem      x86_64 20.1.1-1.fc43                  fedora      15.3 KiB
 cmake-data                     noarch 4.0.0~rc4-2.fc43               copr_base    8.6 MiB
 cmake-filesystem               x86_64 4.0.0~rc4-2.fc43               copr_base    0.0   B
 cmake-rpm-macros               noarch 4.0.0~rc4-2.fc43               copr_base    7.5 KiB
 cpp                            x86_64 15.0.1-0.10.fc43               fedora      37.8 MiB
 default-fonts-core-sans        noarch 4.2-4.fc42                     fedora      11.9 KiB
 emacs-filesystem               noarch 1:30.0-4.fc42                  fedora       0.0   B
 environment-modules            x86_64 5.5.0-3.fc42                   fedora       1.8 MiB
 expat                          x86_64 2.7.0-1.fc43                   fedora     288.8 KiB
 fonts-filesystem               noarch 1:2.0.5-21.fc42                fedora       0.0   B
 gcc                            x86_64 15.0.1-0.10.fc43               fedora     110.7 MiB
 gcc-plugin-annobin             x86_64 15.0.1-0.10.fc43               fedora      57.2 KiB
 git-core                       x86_64 2.49.0-1.fc43                  fedora      22.8 MiB
 git-core-doc                   noarch 2.49.0-1.fc43                  fedora      17.6 MiB
 glibc-devel                    x86_64 2.41.9000-2.fc43               fedora       2.3 MiB
 gnupg2                         x86_64 2.4.7-2.fc42                   fedora       9.8 MiB
 gnutls                         x86_64 3.8.9-5.fc43                   fedora       3.6 MiB
 gnutls-dane                    x86_64 3.8.9-5.fc43                   fedora      69.3 KiB
 google-noto-fonts-common       noarch 20250301-1.fc43                fedora      17.7 KiB
 google-noto-sans-mono-vf-fonts noarch 20250301-1.fc43                fedora     561.2 KiB
 google-noto-sans-vf-fonts      noarch 20250301-1.fc43                fedora       1.4 MiB
 google-noto-serif-vf-fonts     noarch 20250301-1.fc43                fedora       1.6 MiB
 gpgme                          x86_64 1.24.2-1.fc43                  copr_base  587.3 KiB
 groff-base                     x86_64 1.23.0-8.fc42                  fedora       3.9 MiB
 hipblas                        x86_64 6.3.0-5.fc43                   copr_base    1.1 MiB
 hipblas-common-devel           noarch 6.3.0-2.fc43                   copr_base   16.4 KiB
 hipcc                          x86_64 18-43.rocm6.3.2.fc43           fedora     604.5 KiB
 hiredis                        x86_64 1.2.0-6.fc42                   fedora     105.9 KiB
 hwdata                         noarch 0.393-1.fc43                   fedora       9.4 MiB
 hwloc-libs                     x86_64 2.12.0-1.fc43                  fedora       2.9 MiB
 jsoncpp                        x86_64 1.9.6-1.fc43                   copr_base  261.6 KiB
 kernel-headers                 x86_64 6.14.0-0.rc7.56.fc43           fedora       6.5 MiB
 keyutils-libs-devel            x86_64 1.6.3-5.fc42                   fedora      48.2 KiB
 krb5-devel                     x86_64 1.21.3-5.fc42                  fedora     705.9 KiB
 langpacks-core-en              noarch 4.2-4.fc42                     fedora     398.0   B
 langpacks-fonts-en             noarch 4.2-4.fc42                     fedora     341.0   B
 less                           x86_64 668-2.fc42                     fedora     405.8 KiB
 libassuan                      x86_64 2.5.7-3.fc42                   fedora     167.8 KiB
 libb2                          x86_64 0.98.1-13.fc42                 fedora      46.1 KiB
 libcbor                        x86_64 0.11.0-3.fc42                  fedora      77.8 KiB
 libcom_err-devel               x86_64 1.47.2-3.fc42                  fedora      16.7 KiB
 libdrm                         x86_64 2.4.124-2.fc42                 fedora     407.9 KiB
 libedit                        x86_64 3.1-55.20250104cvs.fc42        fedora     244.1 KiB
 libfabric                      x86_64 2.1.0-1.fc43                   fedora       5.4 MiB
 libfido2                       x86_64 1.15.0-3.fc43                  copr_base  238.1 KiB
 libgcrypt                      x86_64 1.11.0-5.fc42                  fedora       1.6 MiB
 libgfortran                    x86_64 15.0.1-0.10.fc43               fedora       3.3 MiB
 libgpg-error                   x86_64 1.51-2.fc42                    fedora     894.1 KiB
 libibverbs                     x86_64 56.0-2.fc43                    fedora       1.2 MiB
 libidn2-devel                  x86_64 2.3.8-1.fc43                   fedora     149.1 KiB
 libkadm5                       x86_64 1.21.3-5.fc42                  fedora     213.9 KiB
 libksba                        x86_64 1.6.7-3.fc42                   fedora     402.5 KiB
 libmpc                         x86_64 1.3.1-7.fc42                   fedora     164.5 KiB
 libnghttp2-devel               x86_64 1.65.0-1.fc43                  fedora     286.3 KiB
 libnl3                         x86_64 3.11.0-3.fc42                  fedora       1.0 MiB
 libomp                         x86_64 20.1.1-1.fc43                  fedora       2.2 MiB
 libomp-devel                   x86_64 20.1.1-1.fc43                  fedora       1.5 MiB
 libpciaccess                   x86_64 0.16-15.fc42                   fedora      44.5 KiB
 libpipeline                    x86_64 1.5.8-2.fc42                   fedora     145.1 KiB
 libpsl-devel                   x86_64 0.21.5-5.fc42                  fedora     110.3 KiB
 libpsm2                        x86_64 12.0.1-2.fc42                  fedora     442.3 KiB
 libquadmath                    x86_64 15.0.1-0.10.fc43               fedora     317.9 KiB
 librdmacm                      x86_64 56.0-2.fc43                    fedora     142.0 KiB
 libselinux-devel               x86_64 3.8-1.fc42                     fedora     126.8 KiB
 libsepol-devel                 x86_64 3.8-1.fc42                     fedora     120.8 KiB
 libssh-devel                   x86_64 0.11.1-4.fc42                  fedora     178.0 KiB
 libstdc++-devel                x86_64 15.0.1-0.10.fc43               fedora      15.9 MiB
 libtommath                     x86_64 1.3.1~rc1-5.fc42               fedora     130.4 KiB
 libusb1                        x86_64 1.0.27-9.fc43                  fedora     166.5 KiB
 libuv                          x86_64 1:1.50.0-1.fc43                copr_base  566.7 KiB
 libverto-devel                 x86_64 0.3.2-10.fc42                  fedora      25.7 KiB
 libxcrypt-devel                x86_64 4.4.38-6.fc43                  fedora      30.8 KiB
 llvm-filesystem                x86_64 20.1.1-1.fc43                  fedora       0.0   B
 llvm-libs                      x86_64 20.1.1-1.fc43                  fedora     137.0 MiB
 make                           x86_64 1:4.4.1-10.fc42                fedora       1.8 MiB
 man-db                         x86_64 2.13.0-2.fc42                  fedora       2.8 MiB
 mpdecimal                      x86_64 4.0.0-2.fc43                   fedora     216.8 KiB
 munge-libs                     x86_64 0.5.16-5.fc43                  fedora      28.0 KiB
 ncurses                        x86_64 6.5-5.20250125.fc42            fedora     608.1 KiB
 nettle                         x86_64 3.10.1-1.fc43                  fedora     790.5 KiB
 npth                           x86_64 1.8-2.fc42                     fedora      49.6 KiB
 numactl-libs                   x86_64 2.0.19-2.fc42                  fedora      52.9 KiB
 openssh                        x86_64 9.9p1-14.fc43                  fedora       1.4 MiB
 openssh-clients                x86_64 9.9p1-14.fc43                  fedora       2.6 MiB
 openssl-devel                  x86_64 1:3.5.0-1.fc43                 fedora       4.6 MiB
 orangefs                       x86_64 2.9.8-14.fc42                  fedora       3.1 MiB
 pcre2-devel                    x86_64 10.45-1.fc43                   fedora       2.1 MiB
 pcre2-utf16                    x86_64 10.45-1.fc43                   fedora     626.3 KiB
 pcre2-utf32                    x86_64 10.45-1.fc43                   fedora     598.2 KiB
 perl-AutoLoader                noarch 5.74-515.fc42                  fedora      20.5 KiB
 perl-B                         x86_64 1.89-515.fc42                  fedora     498.0 KiB
 perl-Carp                      noarch 1.54-512.fc42                  fedora      46.6 KiB
 perl-Class-Struct              noarch 0.68-515.fc42                  fedora      25.4 KiB
 perl-Data-Dumper               x86_64 2.189-513.fc42                 fedora     115.6 KiB
 perl-Digest                    noarch 1.20-512.fc42                  fedora      35.3 KiB
 perl-Digest-MD5                x86_64 2.59-6.fc42                    fedora      59.7 KiB
 perl-DynaLoader                x86_64 1.56-515.fc42                  fedora      32.1 KiB
 perl-Encode                    x86_64 4:3.21-512.fc42                fedora       4.7 MiB
 perl-Errno                     x86_64 1.38-515.fc42                  fedora       8.3 KiB
 perl-Error                     noarch 1:0.17030-1.fc43               fedora      76.7 KiB
 perl-Exporter                  noarch 5.78-512.fc42                  fedora      54.3 KiB
 perl-Fcntl                     x86_64 1.18-515.fc42                  fedora      48.9 KiB
 perl-File-Basename             noarch 2.86-515.fc42                  fedora      14.0 KiB
 perl-File-Copy                 noarch 2.41-515.fc42                  fedora      19.6 KiB
 perl-File-Find                 noarch 1.44-515.fc42                  fedora      41.9 KiB
 perl-File-Path                 noarch 2.18-512.fc42                  fedora      63.5 KiB
 perl-File-Temp                 noarch 1:0.231.100-512.fc42           fedora     162.3 KiB
 perl-File-Which                noarch 1.27-13.fc42                   fedora      30.4 KiB
 perl-File-stat                 noarch 1.14-515.fc42                  fedora      12.5 KiB
 perl-FileHandle                noarch 2.05-515.fc42                  fedora       9.3 KiB
 perl-Getopt-Long               noarch 1:2.58-3.fc42                  fedora     144.5 KiB
 perl-Getopt-Std                noarch 1.14-515.fc42                  fedora      11.2 KiB
 perl-Git                       noarch 2.49.0-1.fc43                  fedora      64.0 KiB
 perl-HTTP-Tiny                 noarch 0.090-2.fc42                   fedora     154.4 KiB
 perl-IO                        x86_64 1.55-515.fc42                  fedora     147.0 KiB
 perl-IO-Socket-IP              noarch 0.43-2.fc42                    fedora     100.3 KiB
 perl-IO-Socket-SSL             noarch 2.089-2.fc42                   fedora     703.3 KiB
 perl-IPC-Open3                 noarch 1.22-515.fc42                  fedora      22.5 KiB
 perl-MIME-Base32               noarch 1.303-23.fc42                  fedora      30.7 KiB
 perl-MIME-Base64               x86_64 3.16-512.fc42                  fedora      42.0 KiB
 perl-Net-SSLeay                x86_64 1.94-8.fc42                    fedora       1.3 MiB
 perl-POSIX                     x86_64 2.20-515.fc42                  fedora     231.0 KiB
 perl-PathTools                 x86_64 3.91-513.fc42                  fedora     180.0 KiB
 perl-Pod-Escapes               noarch 1:1.07-512.fc42                fedora      24.9 KiB
 perl-Pod-Perldoc               noarch 3.28.01-513.fc42               fedora     163.7 KiB
 perl-Pod-Simple                noarch 1:3.45-512.fc42                fedora     560.8 KiB
 perl-Pod-Usage                 noarch 4:2.03-512.fc42                fedora      84.8 KiB
 perl-Scalar-List-Utils         x86_64 5:1.68-2.fc42                  fedora     144.8 KiB
 perl-SelectSaver               noarch 1.02-515.fc42                  fedora       2.2 KiB
 perl-Socket                    x86_64 4:2.038-512.fc42               fedora     119.9 KiB
 perl-Storable                  x86_64 1:3.32-512.fc42                fedora     232.3 KiB
 perl-Symbol                    noarch 1.09-515.fc42                  fedora       6.8 KiB
 perl-Term-ANSIColor            noarch 5.01-513.fc42                  fedora      97.5 KiB
 perl-Term-Cap                  noarch 1.18-512.fc42                  fedora      29.3 KiB
 perl-TermReadKey               x86_64 2.38-24.fc42                   fedora      64.0 KiB
 perl-Text-ParseWords           noarch 3.31-512.fc42                  fedora      13.6 KiB
 perl-Text-Tabs+Wrap            noarch 2024.001-512.fc42              fedora      22.6 KiB
 perl-Time-Local                noarch 2:1.350-512.fc42               fedora      68.9 KiB
 perl-URI                       noarch 5.31-2.fc42                    fedora     257.0 KiB
 perl-base                      noarch 2.27-515.fc42                  fedora      12.5 KiB
 perl-constant                  noarch 1.33-513.fc42                  fedora      26.2 KiB
 perl-if                        noarch 0.61.000-515.fc42              fedora       5.8 KiB
 perl-interpreter               x86_64 4:5.40.1-515.fc42              fedora     118.1 KiB
 perl-lib                       x86_64 0.65-515.fc42                  fedora       8.5 KiB
 perl-libnet                    noarch 3.15-513.fc42                  fedora     289.4 KiB
 perl-libs                      x86_64 4:5.40.1-515.fc42              fedora       9.8 MiB
 perl-locale                    noarch 1.12-515.fc42                  fedora       6.5 KiB
 perl-mro                       x86_64 1.29-515.fc42                  fedora      41.5 KiB
 perl-overload                  noarch 1.37-515.fc42                  fedora      71.5 KiB
 perl-overloading               noarch 0.02-515.fc42                  fedora       4.8 KiB
 perl-parent                    noarch 1:0.244-2.fc42                 fedora      10.3 KiB
 perl-podlators                 noarch 1:6.0.2-3.fc42                 fedora     317.5 KiB
 perl-vars                      noarch 1.05-515.fc42                  fedora       3.9 KiB
 pmix                           x86_64 4.2.8-4.fc42                   fedora       2.0 MiB
 procps-ng                      x86_64 4.0.4-6.fc42                   fedora       1.0 MiB
 protobuf-c                     x86_64 1.5.0-4.fc41                   fedora      54.0 KiB
 prrte                          x86_64 3.0.6-6.fc43                   fedora     158.2 KiB
 prrte-libs                     x86_64 3.0.6-6.fc43                   fedora       1.6 MiB
 pthreadpool                    x86_64 0.0^git20230829.4fe0e1e-6.fc42 fedora     109.5 KiB
 publicsuffix-list              noarch 20250116-1.fc42                fedora     329.8 KiB
 python-pip-wheel               noarch 24.3.1-2.fc42                  fedora       1.2 MiB
 python3                        x86_64 3.13.2-2.fc43                  fedora      27.6 KiB
 python3-libs                   x86_64 3.13.2-2.fc43                  fedora      39.9 MiB
 rhash                          x86_64 1.4.5-2.fc42                   fedora     351.0 KiB
 rocblas                        x86_64 6.3.0-7.fc43                   fedora       3.8 GiB
 rocm-clang                     x86_64 18-43.rocm6.3.2.fc43           fedora      92.3 MiB
 rocm-clang-devel               x86_64 18-43.rocm6.3.2.fc43           fedora      21.8 MiB
 rocm-clang-libs                x86_64 18-43.rocm6.3.2.fc43           fedora      91.0 MiB
 rocm-clang-runtime-devel       x86_64 18-43.rocm6.3.2.fc43           fedora       6.9 MiB
 rocm-comgr                     x86_64 18-43.rocm6.3.2.fc43           fedora     116.3 MiB
 rocm-device-libs               x86_64 18-43.rocm6.3.2.fc43           fedora       3.2 MiB
 rocm-hip                       x86_64 6.3.2-3.fc43                   fedora      23.3 MiB
 rocm-libc++                    x86_64 18-43.rocm6.3.2.fc43           fedora       1.2 MiB
 rocm-libc++-devel              x86_64 18-43.rocm6.3.2.fc43           fedora       7.0 MiB
 rocm-lld                       x86_64 18-43.rocm6.3.2.fc43           fedora       5.3 MiB
 rocm-llvm                      x86_64 18-43.rocm6.3.2.fc43           fedora      68.4 MiB
 rocm-llvm-devel                x86_64 18-43.rocm6.3.2.fc43           fedora      24.3 MiB
 rocm-llvm-filesystem           x86_64 18-43.rocm6.3.2.fc43           fedora       0.0   B
 rocm-llvm-libs                 x86_64 18-43.rocm6.3.2.fc43           fedora      80.7 MiB
 rocm-llvm-static               x86_64 18-43.rocm6.3.2.fc43           fedora     234.4 MiB
 rocm-runtime                   x86_64 6.3.2-5.fc43                   fedora       2.9 MiB
 rocsolver                      x86_64 6.3.0-5.fc43                   fedora     130.2 MiB
 tcl                            x86_64 1:9.0.0-8.fc43                 fedora       4.3 MiB
 tcsh                           x86_64 6.24.14-2.fc42                 fedora       1.2 MiB
 tpm2-tss                       x86_64 4.1.3-6.fc42                   fedora       1.6 MiB
 tzdata                         noarch 2025a-1.fc43                   fedora       1.6 MiB
 ucx                            x86_64 1.17.0-4.fc43                  fedora       2.3 MiB
 unbound-libs                   x86_64 1.22.0-14.fc43                 fedora       1.4 MiB
 vim-filesystem                 noarch 2:9.1.1227-1.fc43              fedora      40.0   B
 wget2                          x86_64 2.2.0-3.fc43                   fedora       1.0 MiB
 wget2-libs                     x86_64 2.2.0-3.fc43                   fedora     365.6 KiB
 zlib-ng-compat-devel           x86_64 2.2.4-2.fc43                   fedora     107.0 KiB

Transaction Summary:
 Installing:       213 packages

Total size of inbound packages is 643 MiB. Need to download 475 MiB.
After this operation, 5 GiB extra will be used (install 5 GiB, remove 0 B).
[  1/213] git-0:2.49.0-1.fc43.x86_64    100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[  2/213] cmake-0:4.0.0~rc4-2.fc43.x86_ 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[  3/213] gcc-c++-0:15.0.1-0.10.fc43.x8 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[  4/213] libcurl-devel-0:8.13.0~rc2-1. 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[  5/213] git-core-0:2.49.0-1.fc43.x86_ 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[  6/213] git-core-doc-0:2.49.0-1.fc43. 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[  7/213] perl-File-Basename-0:2.86-515 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[  8/213] perl-File-Find-0:1.44-515.fc4 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[  9/213] perl-Getopt-Long-1:2.58-3.fc4 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 10/213] perl-Git-0:2.49.0-1.fc43.noar 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 11/213] perl-IPC-Open3-0:1.22-515.fc4 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 12/213] perl-PathTools-0:3.91-513.fc4 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 13/213] perl-TermReadKey-0:2.38-24.fc 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 14/213] perl-interpreter-4:5.40.1-515 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 15/213] perl-lib-0:0.65-515.fc42.x86_ 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 16/213] libgfortran-0:15.0.1-0.10.fc4 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 17/213] libquadmath-0:15.0.1-0.10.fc4 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 18/213] openssh-clients-0:9.9p1-14.fc 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 19/213] perl-File-Copy-0:2.41-515.fc4 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 20/213] perl-Getopt-Std-0:1.14-515.fc 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 21/213] perl-Scalar-List-Utils-5:1.68 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 22/213] perl-URI-0:5.31-2.fc42.noarch 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 23/213] expat-0:2.7.0-1.fc43.x86_64   100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 24/213] make-1:4.4.1-10.fc42.x86_64   100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 25/213] rhash-0:1.4.5-2.fc42.x86_64   100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 26/213] gcc-0:15.0.1-0.10.fc43.x86_64 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 27/213] libmpc-0:1.3.1-7.fc42.x86_64  100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 28/213] less-0:668-2.fc42.x86_64      100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 29/213] perl-Carp-0:1.54-512.fc42.noa 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 30/213] perl-Exporter-0:5.78-512.fc42 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 31/213] perl-Pod-Usage-4:2.03-512.fc4 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 32/213] perl-Text-ParseWords-0:3.31-5 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 33/213] perl-base-0:2.27-515.fc42.noa 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 34/213] perl-constant-0:1.33-513.fc42 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 35/213] perl-overload-0:1.37-515.fc42 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 36/213] perl-Error-1:0.17030-1.fc43.n 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 37/213] perl-Fcntl-0:1.18-515.fc42.x8 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 38/213] perl-IO-0:1.55-515.fc42.x86_6 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 39/213] perl-POSIX-0:2.20-515.fc42.x8 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 40/213] perl-Symbol-0:1.09-515.fc42.n 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 41/213] perl-Errno-0:1.38-515.fc42.x8 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 42/213] perl-libs-4:5.40.1-515.fc42.x 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 43/213] perl-DynaLoader-0:1.56-515.fc 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 44/213] perl-vars-0:1.05-515.fc42.noa 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 45/213] default-fonts-core-sans-0:4.2 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 46/213] libibverbs-0:56.0-2.fc43.x86_ 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 47/213] libnl3-0:3.11.0-3.fc42.x86_64 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 48/213] librdmacm-0:56.0-2.fc43.x86_6 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 49/213] libedit-0:3.1-55.20250104cvs. 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 50/213] openssh-0:9.9p1-14.fc43.x86_6 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 51/213] perl-Data-Dumper-0:2.189-513. 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 52/213] perl-MIME-Base32-0:1.303-23.f 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 53/213] perl-MIME-Base64-0:3.16-512.f 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 54/213] perl-libnet-0:3.15-513.fc42.n 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 55/213] perl-parent-1:0.244-2.fc42.no 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 56/213] vim-filesystem-2:9.1.1227-1.f 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 57/213] libdrm-0:2.4.124-2.fc42.x86_6 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 58/213] cpp-0:15.0.1-0.10.fc43.x86_64 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 59/213] perl-Pod-Perldoc-0:3.28.01-51 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 60/213] perl-podlators-1:6.0.2-3.fc42 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 61/213] perl-mro-0:1.29-515.fc42.x86_ 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 62/213] perl-overloading-0:0.02-515.f 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 63/213] perl-File-stat-0:1.14-515.fc4 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 64/213] perl-SelectSaver-0:1.02-515.f 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 65/213] perl-Socket-4:2.038-512.fc42. 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 66/213] perl-locale-0:1.12-515.fc42.n 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 67/213] abattis-cantarell-vf-fonts-0: 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 68/213] google-noto-sans-vf-fonts-0:2 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 69/213] fonts-filesystem-1:2.0.5-21.f 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 70/213] google-noto-fonts-common-0:20 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 71/213] perl-B-0:1.89-515.fc42.x86_64 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 72/213] perl-Digest-MD5-0:2.59-6.fc42 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 73/213] perl-FileHandle-0:2.05-515.fc 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 74/213] perl-IO-Socket-IP-0:0.43-2.fc 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 75/213] perl-Time-Local-2:1.350-512.f 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 76/213] groff-base-0:1.23.0-8.fc42.x8 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 77/213] libpciaccess-0:0.16-15.fc42.x 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 78/213] gnutls-0:3.8.9-5.fc43.x86_64  100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 79/213] perl-File-Temp-1:0.231.100-51 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 80/213] perl-HTTP-Tiny-0:0.090-2.fc42 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 81/213] perl-Pod-Simple-1:3.45-512.fc 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 82/213] perl-Term-ANSIColor-0:5.01-51 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 83/213] perl-Term-Cap-0:1.18-512.fc42 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 84/213] perl-Class-Struct-0:0.68-515. 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 85/213] perl-if-0:0.61.000-515.fc42.n 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 86/213] perl-Digest-0:1.20-512.fc42.n 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 87/213] hwdata-0:0.393-1.fc43.noarch  100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 88/213] nettle-0:3.10.1-1.fc43.x86_64 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 89/213] perl-File-Path-0:2.18-512.fc4 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 90/213] perl-IO-Socket-SSL-0:2.089-2. 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 91/213] perl-Net-SSLeay-0:1.94-8.fc42 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 92/213] perl-Pod-Escapes-1:1.07-512.f 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 93/213] perl-Text-Tabs+Wrap-0:2024.00 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 94/213] ncurses-0:6.5-5.20250125.fc42 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 95/213] python3-libs-0:3.13.2-2.fc43. 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 96/213] perl-AutoLoader-0:5.74-515.fc 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 97/213] libb2-0:0.98.1-13.fc42.x86_64 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 98/213] mpdecimal-0:4.0.0-2.fc43.x86_ 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[ 99/213] python-pip-wheel-0:24.3.1-2.f 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[100/213] tzdata-0:2025a-1.fc43.noarch  100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[101/213] cmake-filesystem-0:4.0.0~rc4- 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[102/213] zlib-ng-compat-devel-0:2.2.4- 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[103/213] libssh-devel-0:0.11.1-4.fc42. 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[104/213] jsoncpp-0:1.9.6-1.fc43.x86_64 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[105/213] libuv-1:1.50.0-1.fc43.x86_64  100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[106/213] cmake-data-0:4.0.0~rc4-2.fc43 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[107/213] emacs-filesystem-1:30.0-4.fc4 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[108/213] gpgme-0:1.24.2-1.fc43.x86_64  100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[109/213] gnupg2-0:2.4.7-2.fc42.x86_64  100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[110/213] libassuan-0:2.5.7-3.fc42.x86_ 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[111/213] libgpg-error-0:1.51-2.fc42.x8 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[112/213] libgcrypt-0:1.11.0-5.fc42.x86 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[113/213] libksba-0:1.6.7-3.fc42.x86_64 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[114/213] npth-0:1.8-2.fc42.x86_64      100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[115/213] tpm2-tss-0:4.1.3-6.fc42.x86_6 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[116/213] libusb1-0:1.0.27-9.fc43.x86_6 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[117/213] python3-0:3.13.2-2.fc43.x86_6 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[118/213] perl-Encode-4:3.21-512.fc42.x 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[119/213] perl-Storable-1:3.32-512.fc42 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[120/213] libfido2-0:1.15.0-3.fc43.x86_ 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[121/213] libcbor-0:0.11.0-3.fc42.x86_6 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[122/213] brotli-devel-0:1.1.0-6.fc43.x 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[123/213] krb5-devel-0:1.21.3-5.fc42.x8 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[124/213] libkadm5-0:1.21.3-5.fc42.x86_ 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[125/213] libidn2-devel-0:2.3.8-1.fc43. 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[126/213] libnghttp2-devel-0:1.65.0-1.f 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[127/213] libpsl-devel-0:0.21.5-5.fc42. 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[128/213] publicsuffix-list-0:20250116- 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[129/213] openssl-devel-1:3.5.0-1.fc43. 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[130/213] keyutils-libs-devel-0:1.6.3-5 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[131/213] libcom_err-devel-0:1.47.2-3.f 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[132/213] libselinux-devel-0:3.8-1.fc42 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[133/213] libsepol-devel-0:3.8-1.fc42.x 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[134/213] libverto-devel-0:0.3.2-10.fc4 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[135/213] llvm-libs-0:20.1.1-1.fc43.x86 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[136/213] llvm-filesystem-0:20.1.1-1.fc 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[137/213] libstdc++-devel-0:15.0.1-0.10 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[138/213] glibc-devel-0:2.41.9000-2.fc4 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[139/213] libxcrypt-devel-0:4.4.38-6.fc 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[140/213] brotli-0:1.1.0-6.fc43.x86_64  100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[141/213] pcre2-devel-0:10.45-1.fc43.x8 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[142/213] pcre2-utf16-0:10.45-1.fc43.x8 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[143/213] pcre2-utf32-0:10.45-1.fc43.x8 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[144/213] kernel-headers-0:6.14.0-0.rc7 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[145/213] annobin-plugin-gcc-0:12.93-1. 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[146/213] gcc-plugin-annobin-0:15.0.1-0 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[147/213] annobin-docs-0:12.93-1.fc43.n 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[148/213] cmake-rpm-macros-0:4.0.0~rc4- 100% |   0.0   B/s |   0.0   B |  00m00s
>>> Already downloaded                                                          
[149/213] langpacks-en-0:4.2-4.fc42.noa 100% |  73.0 KiB/s |  10.9 KiB |  00m00s
[150/213] hipcc-libomp-devel-0:18-43.ro 100% |  89.6 KiB/s |  13.3 KiB |  00m00s
[151/213] rocm-comgr-devel-0:18-43.rocm 100% | 210.9 KiB/s |  31.6 KiB |  00m00s
[152/213] rocblas-devel-0:6.3.0-7.fc43. 100% | 457.8 KiB/s | 110.3 KiB |  00m00s
[153/213] rocm-rpm-macros-0:6.3.2-2.fc4 100% | 215.8 KiB/s |  15.3 KiB |  00m00s
[154/213] rocm-hip-devel-0:6.3.2-3.fc43 100% | 974.0 KiB/s | 240.6 KiB |  00m00s
[155/213] rocm-runtime-devel-0:6.3.2-5. 100% | 897.2 KiB/s |  92.4 KiB |  00m00s
[156/213] wget2-wget-0:2.2.0-3.fc43.x86 100% | 141.4 KiB/s |   9.8 KiB |  00m00s
[157/213] xxd-2:9.1.1227-1.fc43.x86_64  100% | 386.9 KiB/s |  32.1 KiB |  00m00s
[158/213] openmpi-0:5.0.6-5.fc43.x86_64 100% |   3.0 MiB/s |   2.0 MiB |  00m01s
[159/213] hipblas-devel-0:6.3.0-5.fc43. 100% |   1.6 MiB/s | 105.3 KiB |  00m00s
[160/213] pthreadpool-devel-0:0.0^git20 100% | 200.2 KiB/s |  14.4 KiB |  00m00s
[161/213] langpacks-core-en-0:4.2-4.fc4 100% | 165.2 KiB/s |  10.9 KiB |  00m00s
[162/213] hipcc-0:18-43.rocm6.3.2.fc43. 100% |   1.4 MiB/s | 126.1 KiB |  00m00s
[163/213] langpacks-fonts-en-0:4.2-4.fc 100% | 147.6 KiB/s |  11.2 KiB |  00m00s
[164/213] hwloc-libs-0:2.12.0-1.fc43.x8 100% |  15.2 MiB/s |   2.1 MiB |  00m00s
[165/213] libpsm2-0:12.0.1-2.fc42.x86_6 100% |   1.3 MiB/s | 202.8 KiB |  00m00s
[166/213] orangefs-0:2.9.8-14.fc42.x86_ 100% |  18.9 MiB/s |   1.9 MiB |  00m00s
[167/213] libfabric-0:2.1.0-1.fc43.x86_ 100% |   5.8 MiB/s |   1.5 MiB |  00m00s
[168/213] prrte-0:3.0.6-6.fc43.x86_64   100% | 821.5 KiB/s |  55.9 KiB |  00m00s
[169/213] ucx-0:1.17.0-4.fc43.x86_64    100% |   8.3 MiB/s | 827.6 KiB |  00m00s
[170/213] pmix-0:4.2.8-4.fc42.x86_64    100% |   3.4 MiB/s | 677.0 KiB |  00m00s
[171/213] rocm-device-libs-0:18-43.rocm 100% |   4.6 MiB/s | 500.7 KiB |  00m00s
[172/213] perl-File-Which-0:1.27-13.fc4 100% | 313.2 KiB/s |  21.6 KiB |  00m00s
[173/213] rocm-hip-0:6.3.2-3.fc43.x86_6 100% |  14.8 MiB/s |   9.3 MiB |  00m01s
[174/213] rocm-comgr-0:18-43.rocm6.3.2. 100% |  30.5 MiB/s |  29.1 MiB |  00m01s
[175/213] environment-modules-0:5.5.0-3 100% |   6.0 MiB/s | 764.7 KiB |  00m00s
[176/213] wget2-0:2.2.0-3.fc43.x86_64   100% |   3.7 MiB/s | 279.3 KiB |  00m00s
[177/213] rocm-runtime-0:6.3.2-5.fc43.x 100% |   6.2 MiB/s | 615.6 KiB |  00m00s
[178/213] pthreadpool-0:0.0^git20230829 100% | 674.9 KiB/s |  46.6 KiB |  00m00s
[179/213] google-noto-sans-mono-vf-font 100% |   3.4 MiB/s | 276.9 KiB |  00m00s
[180/213] google-noto-serif-vf-fonts-0: 100% |   8.1 MiB/s | 665.5 KiB |  00m00s
[181/213] numactl-libs-0:2.0.19-2.fc42. 100% | 434.4 KiB/s |  31.3 KiB |  00m00s
[182/213] tcsh-0:6.24.14-2.fc42.x86_64  100% |   6.0 MiB/s | 462.8 KiB |  00m00s
[183/213] munge-libs-0:0.5.16-5.fc43.x8 100% | 296.6 KiB/s |  20.5 KiB |  00m00s
[184/213] prrte-libs-0:3.0.6-6.fc43.x86 100% |   7.0 MiB/s | 545.8 KiB |  00m00s
[185/213] rocm-lld-0:18-43.rocm6.3.2.fc 100% |  14.7 MiB/s |   1.4 MiB |  00m00s
[186/213] rocm-clang-devel-0:18-43.rocm 100% |  15.4 MiB/s |   2.4 MiB |  00m00s
[187/213] man-db-0:2.13.0-2.fc42.x86_64 100% |  10.7 MiB/s |   1.3 MiB |  00m00s
[188/213] wget2-libs-0:2.2.0-3.fc43.x86 100% |   1.8 MiB/s | 146.4 KiB |  00m00s
[189/213] rocm-llvm-static-0:18-43.rocm 100% |  41.7 MiB/s |  27.5 MiB |  00m01s
[190/213] rocm-clang-0:18-43.rocm6.3.2. 100% |  22.3 MiB/s |  20.9 MiB |  00m01s
[191/213] rocm-clang-libs-0:18-43.rocm6 100% |  35.2 MiB/s |  21.4 MiB |  00m01s
[192/213] rocm-llvm-devel-0:18-43.rocm6 100% |  24.4 MiB/s |   3.9 MiB |  00m00s
[193/213] libpipeline-0:1.5.8-2.fc42.x8 100% | 869.9 KiB/s |  60.0 KiB |  00m00s
[194/213] gnutls-dane-0:3.8.9-5.fc43.x8 100% | 535.6 KiB/s |  42.3 KiB |  00m00s
[195/213] rocm-clang-runtime-devel-0:18 100% |   6.6 MiB/s | 523.9 KiB |  00m00s
[196/213] rocm-libc++-devel-0:18-43.roc 100% |   4.2 MiB/s | 993.4 KiB |  00m00s
[197/213] rocm-libc++-0:18-43.rocm6.3.2 100% |   2.1 MiB/s | 341.9 KiB |  00m00s
[198/213] rocm-llvm-filesystem-0:18-43. 100% | 324.9 KiB/s |  21.4 KiB |  00m00s
[199/213] rocm-llvm-libs-0:18-43.rocm6. 100% |  15.9 MiB/s |  19.3 MiB |  00m01s
[200/213] unbound-libs-0:1.22.0-14.fc43 100% |   5.7 MiB/s | 559.1 KiB |  00m00s
[201/213] hiredis-0:1.2.0-6.fc42.x86_64 100% | 757.2 KiB/s |  50.7 KiB |  00m00s
[202/213] protobuf-c-0:1.5.0-4.fc41.x86 100% | 483.5 KiB/s |  32.4 KiB |  00m00s
[203/213] hipblas-0:6.3.0-5.fc43.x86_64 100% |   7.5 MiB/s | 161.2 KiB |  00m00s
[204/213] rocm-llvm-0:18-43.rocm6.3.2.f 100% |  23.8 MiB/s |  15.8 MiB |  00m01s
[205/213] hipblas-common-devel-0:6.3.0- 100% | 318.8 KiB/s |  13.4 KiB |  00m00s
[206/213] libomp-devel-0:20.1.1-1.fc43. 100% |   3.7 MiB/s | 281.2 KiB |  00m00s
[207/213] clang-resource-filesystem-0:2 100% | 303.4 KiB/s |  20.0 KiB |  00m00s
[208/213] libomp-0:20.1.1-1.fc43.x86_64 100% |   8.3 MiB/s | 736.7 KiB |  00m00s
[209/213] procps-ng-0:4.0.4-6.fc42.x86_ 100% |   4.6 MiB/s | 365.3 KiB |  00m00s
[210/213] tcl-1:9.0.0-8.fc43.x86_64     100% |  11.9 MiB/s |   1.2 MiB |  00m00s
[211/213] libtommath-0:1.3.1~rc1-5.fc42 100% | 975.3 KiB/s |  64.4 KiB |  00m00s
[212/213] rocblas-0:6.3.0-7.fc43.x86_64 100% |  29.9 MiB/s | 192.7 MiB |  00m06s
[213/213] rocsolver-0:6.3.0-5.fc43.x86_ 100% |  25.3 MiB/s | 109.9 MiB |  00m04s
--------------------------------------------------------------------------------
[213/213] Total                         100% |  48.7 MiB/s | 474.8 MiB |  00m10s
Running transaction
[  1/215] Verify package files          100% | 106.0   B/s | 213.0   B |  00m02s
[  2/215] Prepare transaction           100% | 750.0   B/s | 213.0   B |  00m00s
[  3/215] Installing cmake-filesystem-0 100% |   2.4 MiB/s |   7.5 KiB |  00m00s
[  4/215] Installing libgpg-error-0:1.5 100% |  41.9 MiB/s | 900.0 KiB |  00m00s
[  5/215] Installing fonts-filesystem-1 100% | 769.5 KiB/s | 788.0   B |  00m00s
[  6/215] Installing numactl-libs-0:2.0 100% |  52.5 MiB/s |  53.8 KiB |  00m00s
[  7/215] Installing hwloc-libs-0:2.12. 100% | 360.3 MiB/s |   2.9 MiB |  00m00s
[  8/215] Installing google-noto-fonts- 100% |  18.1 MiB/s |  18.5 KiB |  00m00s
[  9/215] Installing libnl3-0:3.11.0-3. 100% | 170.6 MiB/s |   1.0 MiB |  00m00s
[ 10/215] Installing libibverbs-0:56.0- 100% | 129.9 MiB/s |   1.2 MiB |  00m00s
[ 11/215] Installing less-0:668-2.fc42. 100% |  25.0 MiB/s | 409.1 KiB |  00m00s
[ 12/215] Installing libmpc-0:1.3.1-7.f 100% |  81.1 MiB/s | 166.1 KiB |  00m00s
[ 13/215] Installing expat-0:2.7.0-1.fc 100% |  18.9 MiB/s | 290.8 KiB |  00m00s
[ 14/215] Installing libpsm2-0:12.0.1-2 100% | 144.3 MiB/s | 443.4 KiB |  00m00s
[ 15/215] Installing libassuan-0:2.5.7- 100% |  82.8 MiB/s | 169.6 KiB |  00m00s
[ 16/215] Installing zlib-ng-compat-dev 100% | 106.0 MiB/s | 108.5 KiB |  00m00s
[ 17/215] Installing nettle-0:3.10.1-1. 100% | 193.8 MiB/s | 793.6 KiB |  00m00s
[ 18/215] Installing gnutls-0:3.8.9-5.f 100% | 255.3 MiB/s |   3.6 MiB |  00m00s
[ 19/215] Installing rocm-llvm-filesyst 100% |   2.3 MiB/s |  14.3 KiB |  00m00s
[ 20/215] Installing rocm-libc++-0:18-4 100% |  57.6 MiB/s |   1.2 MiB |  00m00s
[ 21/215] Installing rocm-llvm-libs-0:1 100% |  67.4 MiB/s |  80.7 MiB |  00m01s
[ 22/215] Installing rocm-clang-libs-0: 100% |  67.7 MiB/s |  91.0 MiB |  00m01s
[ 23/215] Installing groff-base-0:1.23. 100% |  74.9 MiB/s |   3.9 MiB |  00m00s
[ 24/215] Installing vim-filesystem-2:9 100% |   2.3 MiB/s |   4.7 KiB |  00m00s
[ 25/215] Installing libedit-0:3.1-55.2 100% | 120.0 MiB/s | 245.8 KiB |  00m00s
[ 26/215] Installing make-1:4.4.1-10.fc 100% |  78.3 MiB/s |   1.8 MiB |  00m00s
[ 27/215] Installing rocm-comgr-0:18-43 100% |  65.4 MiB/s | 116.3 MiB |  00m02s
[ 28/215] Installing rocm-lld-0:18-43.r 100% |  59.8 MiB/s |   5.3 MiB |  00m00s
[ 29/215] Installing rocm-libc++-devel- 100% |  65.5 MiB/s |   7.2 MiB |  00m00s
[ 30/215] Installing cpp-0:15.0.1-0.10. 100% | 286.1 MiB/s |  37.8 MiB |  00m00s
[ 31/215] Installing librdmacm-0:56.0-2 100% |  70.2 MiB/s | 143.8 KiB |  00m00s
[ 32/215] Installing libfabric-0:2.1.0- 100% | 186.4 MiB/s |   5.4 MiB |  00m00s
[ 33/215] Installing google-noto-sans-m 100% | 183.0 MiB/s | 562.2 KiB |  00m00s
[ 34/215] Installing google-noto-serif- 100% | 265.0 MiB/s |   1.6 MiB |  00m00s
[ 35/215] Installing google-noto-sans-v 100% | 278.3 MiB/s |   1.4 MiB |  00m00s
[ 36/215] Installing abattis-cantarell- 100% |  94.9 MiB/s | 194.4 KiB |  00m00s
[ 37/215] Installing default-fonts-core 100% |   8.9 MiB/s |  18.2 KiB |  00m00s
[ 38/215] Installing langpacks-core-en- 100% |   0.0   B/s | 704.0   B |  00m00s
[ 39/215] Installing langpacks-fonts-en 100% |   0.0   B/s | 652.0   B |  00m00s
[ 40/215] Installing libgcrypt-0:1.11.0 100% | 261.5 MiB/s |   1.6 MiB |  00m00s
[ 41/215] Installing libksba-0:1.6.7-3. 100% | 131.9 MiB/s | 405.1 KiB |  00m00s
[ 42/215] Installing libssh-devel-0:0.1 100% | 176.3 MiB/s | 180.5 KiB |  00m00s
[ 43/215] Installing hipblas-common-dev 100% |  17.4 MiB/s |  17.8 KiB |  00m00s
[ 44/215] Installing annobin-docs-0:12. 100% |  32.6 MiB/s | 100.0 KiB |  00m00s
[ 45/215] Installing kernel-headers-0:6 100% | 113.1 MiB/s |   6.7 MiB |  00m00s
[ 46/215] Installing libxcrypt-devel-0: 100% |  10.8 MiB/s |  33.1 KiB |  00m00s
[ 47/215] Installing glibc-devel-0:2.41 100% |  83.3 MiB/s |   2.3 MiB |  00m00s
[ 48/215] Installing gcc-0:15.0.1-0.10. 100% | 311.9 MiB/s | 110.7 MiB |  00m00s
[ 49/215] Installing pcre2-utf32-0:10.4 100% | 195.0 MiB/s | 599.1 KiB |  00m00s
[ 50/215] Installing pcre2-utf16-0:10.4 100% | 204.2 MiB/s | 627.2 KiB |  00m00s
[ 51/215] Installing pcre2-devel-0:10.4 100% |  63.4 MiB/s |   2.1 MiB |  00m00s
[ 52/215] Installing brotli-0:1.1.0-6.f 100% |   2.3 MiB/s |  32.3 KiB |  00m00s
[ 53/215] Installing brotli-devel-0:1.1 100% |  66.4 MiB/s |  68.0 KiB |  00m00s
[ 54/215] Installing libtommath-0:1.3.1 100% |  64.2 MiB/s | 131.5 KiB |  00m00s
[ 55/215] Installing tcl-1:9.0.0-8.fc43 100% | 117.1 MiB/s |   4.3 MiB |  00m00s
[ 56/215] Installing procps-ng-0:4.0.4- 100% |  43.9 MiB/s |   1.0 MiB |  00m00s
[ 57/215] Installing libstdc++-devel-0: 100% | 216.7 MiB/s |  16.0 MiB |  00m00s
[ 58/215] Installing llvm-filesystem-0: 100% |   1.0 MiB/s |   1.1 KiB |  00m00s
[ 59/215] Installing llvm-libs-0:20.1.1 100% | 356.7 MiB/s | 137.0 MiB |  00m00s
[ 60/215] Installing libomp-0:20.1.1-1. 100% | 277.2 MiB/s |   2.2 MiB |  00m00s
[ 61/215] Installing clang-resource-fil 100% |  16.3 MiB/s |  16.7 KiB |  00m00s
[ 62/215] Installing libomp-devel-0:20. 100% | 309.3 MiB/s |   1.5 MiB |  00m00s
[ 63/215] Installing libverto-devel-0:0 100% |  25.7 MiB/s |  26.4 KiB |  00m00s
[ 64/215] Installing libsepol-devel-0:3 100% |  31.3 MiB/s | 128.3 KiB |  00m00s
[ 65/215] Installing libselinux-devel-0 100% |  17.5 MiB/s | 161.6 KiB |  00m00s
[ 66/215] Installing libcom_err-devel-0 100% |   1.3 MiB/s |  18.3 KiB |  00m00s
[ 67/215] Installing keyutils-libs-deve 100% |   5.4 MiB/s |  55.2 KiB |  00m00s
[ 68/215] Installing openssl-devel-1:3. 100% |  29.3 MiB/s |   5.6 MiB |  00m00s
[ 69/215] Installing publicsuffix-list- 100% | 161.5 MiB/s | 330.8 KiB |  00m00s
[ 70/215] Installing libpsl-devel-0:0.2 100% |  55.5 MiB/s | 113.6 KiB |  00m00s
[ 71/215] Installing libnghttp2-devel-0 100% | 280.7 MiB/s | 287.5 KiB |  00m00s
[ 72/215] Installing libidn2-devel-0:2. 100% |  51.0 MiB/s | 156.7 KiB |  00m00s
[ 73/215] Installing libkadm5-0:1.21.3- 100% | 105.4 MiB/s | 215.9 KiB |  00m00s
[ 74/215] Installing krb5-devel-0:1.21. 100% |  38.8 MiB/s | 715.2 KiB |  00m00s
[ 75/215] Installing libcbor-0:0.11.0-3 100% |  77.3 MiB/s |  79.2 KiB |  00m00s
[ 76/215] Installing libfido2-0:1.15.0- 100% | 117.0 MiB/s | 239.6 KiB |  00m00s
[ 77/215] Installing libusb1-0:1.0.27-9 100% |   5.5 MiB/s | 168.2 KiB |  00m00s
>>> Running unknown scriptlet: tpm2-tss-0:4.1.3-6.fc42.x86_64                   
>>> Finished unknown scriptlet: tpm2-tss-0:4.1.3-6.fc42.x86_64                  
>>> Scriptlet output:                                                           
>>> Creating group 'tss' with GID 59.                                           
>>> Creating user 'tss' (Account used for TPM access) with UID 59 and GID 59.   
>>>                                                                             
[ 78/215] Installing tpm2-tss-0:4.1.3-6 100% | 156.8 MiB/s |   1.6 MiB |  00m00s
[ 79/215] Installing npth-0:1.8-2.fc42. 100% |  49.5 MiB/s |  50.7 KiB |  00m00s
[ 80/215] Installing gnupg2-0:2.4.7-2.f 100% | 175.1 MiB/s |   9.8 MiB |  00m00s
[ 81/215] Installing gpgme-0:1.24.2-1.f 100% |  33.9 MiB/s | 589.9 KiB |  00m00s
[ 82/215] Installing emacs-filesystem-1 100% |   0.0   B/s | 544.0   B |  00m00s
[ 83/215] Installing libuv-1:1.50.0-1.f 100% | 185.4 MiB/s | 569.5 KiB |  00m00s
[ 84/215] Installing jsoncpp-0:1.9.6-1. 100% |  42.8 MiB/s | 263.1 KiB |  00m00s
[ 85/215] Installing tzdata-0:2025a-1.f 100% |  25.5 MiB/s |   1.9 MiB |  00m00s
[ 86/215] Installing python-pip-wheel-0 100% | 414.7 MiB/s |   1.2 MiB |  00m00s
[ 87/215] Installing mpdecimal-0:4.0.0- 100% | 106.6 MiB/s | 218.4 KiB |  00m00s
[ 88/215] Installing libb2-0:0.98.1-13. 100% |   6.6 MiB/s |  47.2 KiB |  00m00s
[ 89/215] Installing python3-libs-0:3.1 100% | 212.0 MiB/s |  40.3 MiB |  00m00s
[ 90/215] Installing python3-0:3.13.2-2 100% |   1.9 MiB/s |  29.4 KiB |  00m00s
[ 91/215] Installing cmake-rpm-macros-0 100% |   8.0 MiB/s |   8.2 KiB |  00m00s
[ 92/215] Installing rocm-llvm-0:18-43. 100% |  66.7 MiB/s |  68.4 MiB |  00m01s
[ 93/215] Installing rocm-llvm-devel-0: 100% |  70.5 MiB/s |  24.7 MiB |  00m00s
[ 94/215] Installing rocm-llvm-static-0 100% |  93.2 MiB/s | 234.4 MiB |  00m03s
[ 95/215] Installing protobuf-c-0:1.5.0 100% |  27.1 MiB/s |  55.5 KiB |  00m00s
[ 96/215] Installing hiredis-0:1.2.0-6. 100% |   3.8 MiB/s | 107.6 KiB |  00m00s
>>> Running unknown scriptlet: unbound-libs-0:1.22.0-14.fc43.x86_64             
>>> Finished unknown scriptlet: unbound-libs-0:1.22.0-14.fc43.x86_64            
>>> Scriptlet output:                                                           
>>> Creating group 'unbound' with GID 999.                                      
>>> Creating user 'unbound' (Unbound DNS resolver) with UID 999 and GID 999.    
>>>                                                                             
[ 97/215] Installing unbound-libs-0:1.2 100% | 180.0 MiB/s |   1.4 MiB |  00m00s
[ 98/215] Installing gnutls-dane-0:3.8. 100% |  68.5 MiB/s |  70.2 KiB |  00m00s
[ 99/215] Installing wget2-libs-0:2.2.0 100% | 119.4 MiB/s | 366.8 KiB |  00m00s
[100/215] Installing wget2-0:2.2.0-3.fc 100% |  47.8 MiB/s |   1.1 MiB |  00m00s
[101/215] Installing ncurses-0:6.5-5.20 100% |  33.3 MiB/s | 614.7 KiB |  00m00s
[102/215] Installing perl-Digest-0:1.20 100% |  36.2 MiB/s |  37.1 KiB |  00m00s
[103/215] Installing perl-FileHandle-0: 100% |   9.5 MiB/s |   9.8 KiB |  00m00s
[104/215] Installing perl-B-0:1.89-515. 100% | 163.2 MiB/s | 501.3 KiB |  00m00s
[105/215] Installing perl-Digest-MD5-0: 100% |  60.1 MiB/s |  61.6 KiB |  00m00s
[106/215] Installing perl-MIME-Base32-0 100% |  31.4 MiB/s |  32.2 KiB |  00m00s
[107/215] Installing perl-Data-Dumper-0 100% |  57.4 MiB/s | 117.5 KiB |  00m00s
[108/215] Installing perl-libnet-0:3.15 100% |  95.9 MiB/s | 294.7 KiB |  00m00s
[109/215] Installing perl-IO-Socket-IP- 100% |  49.9 MiB/s | 102.2 KiB |  00m00s
[110/215] Installing perl-URI-0:5.31-2. 100% |  52.7 MiB/s | 269.6 KiB |  00m00s
[111/215] Installing perl-AutoLoader-0: 100% |   0.0   B/s |  20.9 KiB |  00m00s
[112/215] Installing perl-locale-0:1.12 100% |   0.0   B/s |   6.9 KiB |  00m00s
[113/215] Installing perl-Time-Local-2: 100% |  68.9 MiB/s |  70.6 KiB |  00m00s
[114/215] Installing perl-if-0:0.61.000 100% |   0.0   B/s |   6.2 KiB |  00m00s
[115/215] Installing perl-File-Path-0:2 100% |  63.0 MiB/s |  64.5 KiB |  00m00s
[116/215] Installing perl-Pod-Escapes-1 100% |  25.3 MiB/s |  25.9 KiB |  00m00s
[117/215] Installing perl-Text-Tabs+Wra 100% |  23.3 MiB/s |  23.9 KiB |  00m00s
[118/215] Installing perl-IO-Socket-SSL 100% | 172.7 MiB/s | 707.4 KiB |  00m00s
[119/215] Installing perl-Net-SSLeay-0: 100% | 151.0 MiB/s |   1.4 MiB |  00m00s
[120/215] Installing perl-POSIX-0:2.20- 100% | 113.4 MiB/s | 232.2 KiB |  00m00s
[121/215] Installing perl-Term-ANSIColo 100% |  96.9 MiB/s |  99.2 KiB |  00m00s
[122/215] Installing perl-Term-Cap-0:1. 100% |  29.9 MiB/s |  30.6 KiB |  00m00s
[123/215] Installing perl-IPC-Open3-0:1 100% |  22.7 MiB/s |  23.3 KiB |  00m00s
[124/215] Installing perl-Class-Struct- 100% |  25.3 MiB/s |  25.9 KiB |  00m00s
[125/215] Installing perl-File-Temp-1:0 100% |  80.1 MiB/s | 164.1 KiB |  00m00s
[126/215] Installing perl-HTTP-Tiny-0:0 100% |  76.4 MiB/s | 156.4 KiB |  00m00s
[127/215] Installing perl-Pod-Simple-1: 100% | 139.3 MiB/s | 570.4 KiB |  00m00s
[128/215] Installing perl-Symbol-0:1.09 100% |   0.0   B/s |   7.2 KiB |  00m00s
[129/215] Installing perl-SelectSaver-0 100% |   0.0   B/s |   2.6 KiB |  00m00s
[130/215] Installing perl-Socket-4:2.03 100% |  59.6 MiB/s | 122.0 KiB |  00m00s
[131/215] Installing perl-File-stat-0:1 100% |  12.7 MiB/s |  13.1 KiB |  00m00s
[132/215] Installing perl-Pod-Perldoc-0 100% |  10.3 MiB/s | 169.2 KiB |  00m00s
[133/215] Installing perl-podlators-1:6 100% |  19.6 MiB/s | 321.4 KiB |  00m00s
[134/215] Installing perl-Text-ParseWor 100% |  14.2 MiB/s |  14.6 KiB |  00m00s
[135/215] Installing perl-base-0:2.27-5 100% |   0.0   B/s |  12.9 KiB |  00m00s
[136/215] Installing perl-Fcntl-0:1.18- 100% |  48.9 MiB/s |  50.0 KiB |  00m00s
[137/215] Installing perl-mro-0:1.29-51 100% |  41.6 MiB/s |  42.6 KiB |  00m00s
[138/215] Installing perl-overloading-0 100% |   5.4 MiB/s |   5.5 KiB |  00m00s
[139/215] Installing perl-IO-0:1.55-515 100% |  73.8 MiB/s | 151.1 KiB |  00m00s
[140/215] Installing perl-Pod-Usage-4:2 100% |   6.0 MiB/s |  86.3 KiB |  00m00s
[141/215] Installing perl-Getopt-Std-0: 100% |  11.5 MiB/s |  11.7 KiB |  00m00s
[142/215] Installing perl-Scalar-List-U 100% |  48.4 MiB/s | 148.5 KiB |  00m00s
[143/215] Installing perl-constant-0:1. 100% |  26.7 MiB/s |  27.4 KiB |  00m00s
[144/215] Installing perl-Errno-0:1.38- 100% |   0.0   B/s |   8.7 KiB |  00m00s
[145/215] Installing perl-vars-0:1.05-5 100% |   0.0   B/s |   4.3 KiB |  00m00s
[146/215] Installing perl-MIME-Base64-0 100% |   3.9 MiB/s |  44.3 KiB |  00m00s
[147/215] Installing perl-parent-1:0.24 100% |  10.7 MiB/s |  11.0 KiB |  00m00s
[148/215] Installing perl-overload-0:1. 100% |  70.3 MiB/s |  71.9 KiB |  00m00s
[149/215] Installing perl-Storable-1:3. 100% | 114.2 MiB/s | 233.9 KiB |  00m00s
[150/215] Installing perl-Getopt-Long-1 100% |  71.9 MiB/s | 147.2 KiB |  00m00s
[151/215] Installing perl-File-Basename 100% |   0.0   B/s |  14.6 KiB |  00m00s
[152/215] Installing perl-Carp-0:1.54-5 100% |  46.6 MiB/s |  47.7 KiB |  00m00s
[153/215] Installing perl-Exporter-0:5. 100% |  54.3 MiB/s |  55.6 KiB |  00m00s
[154/215] Installing perl-PathTools-0:3 100% |  60.1 MiB/s | 184.5 KiB |  00m00s
[155/215] Installing perl-DynaLoader-0: 100% |  31.7 MiB/s |  32.5 KiB |  00m00s
[156/215] Installing perl-Encode-4:3.21 100% | 138.1 MiB/s |   4.7 MiB |  00m00s
[157/215] Installing perl-libs-4:5.40.1 100% | 152.4 MiB/s |   9.9 MiB |  00m00s
[158/215] Installing perl-interpreter-4 100% |   7.8 MiB/s | 119.8 KiB |  00m00s
[159/215] Installing perl-File-Find-0:1 100% |  41.5 MiB/s |  42.5 KiB |  00m00s
[160/215] Installing perl-TermReadKey-0 100% |  32.3 MiB/s |  66.2 KiB |  00m00s
[161/215] Installing perl-lib-0:0.65-51 100% |   0.0   B/s |   8.9 KiB |  00m00s
[162/215] Installing perl-File-Copy-0:2 100% |   0.0   B/s |  20.2 KiB |  00m00s
[163/215] Installing perl-File-Which-0: 100% |  30.7 MiB/s |  31.4 KiB |  00m00s
[164/215] Installing perl-Error-1:0.170 100% |  39.0 MiB/s |  80.0 KiB |  00m00s
[165/215] Installing hwdata-0:0.393-1.f 100% | 428.7 MiB/s |   9.4 MiB |  00m00s
[166/215] Installing libpciaccess-0:0.1 100% |  44.8 MiB/s |  45.9 KiB |  00m00s
[167/215] Installing libdrm-0:2.4.124-2 100% | 134.1 MiB/s | 411.8 KiB |  00m00s
[168/215] Installing rocm-runtime-0:6.3 100% | 363.8 MiB/s |   2.9 MiB |  00m00s
[169/215] Installing rocm-runtime-devel 100% | 185.3 MiB/s | 569.2 KiB |  00m00s
[170/215] Installing rocm-clang-runtime 100% | 106.3 MiB/s |   6.9 MiB |  00m00s
[171/215] Installing libpipeline-0:1.5. 100% |   6.8 MiB/s | 146.6 KiB |  00m00s
[172/215] Installing man-db-0:2.13.0-2. 100% |  50.0 MiB/s |   2.8 MiB |  00m00s
[173/215] Installing environment-module 100% |  42.0 MiB/s |   1.8 MiB |  00m00s
[174/215] Installing munge-libs-0:0.5.1 100% |  28.1 MiB/s |  28.8 KiB |  00m00s
[175/215] Installing pmix-0:4.2.8-4.fc4 100% | 202.7 MiB/s |   2.0 MiB |  00m00s
[176/215] Installing prrte-libs-0:3.0.6 100% | 165.7 MiB/s |   1.7 MiB |  00m00s
[177/215] Installing prrte-0:3.0.6-6.fc 100% |  79.2 MiB/s | 162.1 KiB |  00m00s
[178/215] Installing tcsh-0:6.24.14-2.f 100% |  33.0 MiB/s |   1.3 MiB |  00m00s
[179/215] Installing orangefs-0:2.9.8-1 100% | 100.7 MiB/s |   3.1 MiB |  00m00s
[180/215] Installing openssh-0:9.9p1-14 100% |  69.1 MiB/s |   1.4 MiB |  00m00s
[181/215] Installing openssh-clients-0: 100% |  74.5 MiB/s |   2.6 MiB |  00m00s
[182/215] Installing git-core-0:2.49.0- 100% | 262.1 MiB/s |  22.8 MiB |  00m00s
[183/215] Installing git-core-doc-0:2.4 100% | 191.2 MiB/s |  17.8 MiB |  00m00s
[184/215] Installing perl-Git-0:2.49.0- 100% |  63.5 MiB/s |  65.0 KiB |  00m00s
[185/215] Installing git-0:2.49.0-1.fc4 100% |  85.4 MiB/s |  87.5 KiB |  00m00s
[186/215] Installing rocm-clang-0:18-43 100% |  70.0 MiB/s |  92.3 MiB |  00m01s
[187/215] Installing rocm-clang-devel-0 100% |  93.4 MiB/s |  21.9 MiB |  00m00s
[188/215] Installing rocm-device-libs-0 100% |  78.9 MiB/s |   3.2 MiB |  00m00s
[189/215] Installing hipcc-0:18-43.rocm 100% |  26.9 MiB/s | 605.9 KiB |  00m00s
[190/215] Installing rocm-hip-0:6.3.2-3 100% | 291.2 MiB/s |  23.3 MiB |  00m00s
[191/215] Installing rocblas-0:6.3.0-7. 100% | 146.7 MiB/s |   3.8 GiB |  00m26s
[192/215] Installing rocsolver-0:6.3.0- 100% |  43.5 MiB/s | 130.2 MiB |  00m03s
[193/215] Installing hipblas-0:6.3.0-5. 100% |  80.9 MiB/s |   1.1 MiB |  00m00s
[194/215] Installing rocm-comgr-devel-0 100% |  51.0 MiB/s | 104.4 KiB |  00m00s
[195/215] Installing pthreadpool-0:0.0^ 100% | 107.9 MiB/s | 110.5 KiB |  00m00s
[196/215] Installing rhash-0:1.4.5-2.fc 100% |  15.8 MiB/s | 356.4 KiB |  00m00s
[197/215] Installing cmake-data-0:4.0.0 100% |  54.4 MiB/s |   9.2 MiB |  00m00s
[198/215] Installing cmake-0:4.0.0~rc4- 100% | 271.3 MiB/s |  34.5 MiB |  00m00s
[199/215] Installing ucx-0:1.17.0-4.fc4 100% |  98.2 MiB/s |   2.4 MiB |  00m00s
[200/215] Installing libquadmath-0:15.0 100% | 155.8 MiB/s | 319.2 KiB |  00m00s
[201/215] Installing libgfortran-0:15.0 100% | 301.1 MiB/s |   3.3 MiB |  00m00s
[202/215] Installing openmpi-0:5.0.6-5. 100% | 268.7 MiB/s |   7.0 MiB |  00m00s
[203/215] Installing pthreadpool-devel- 100% |  97.5 MiB/s |  99.8 KiB |  00m00s
[204/215] Installing rocm-hip-devel-0:6 100% | 112.0 MiB/s |   2.7 MiB |  00m00s
[205/215] Installing hipblas-devel-0:6. 100% | 155.7 MiB/s |   3.1 MiB |  00m00s
[206/215] Installing rocblas-devel-0:6. 100% | 139.6 MiB/s |   2.8 MiB |  00m00s
[207/215] Installing hipcc-libomp-devel 100% |   0.0   B/s | 124.0   B |  00m00s
[208/215] Installing rocm-rpm-macros-0: 100% |  19.0 MiB/s |  19.4 KiB |  00m00s
[209/215] Installing wget2-wget-0:2.2.0 100% |  28.9 KiB/s | 444.0   B |  00m00s
[210/215] Installing libcurl-devel-0:8. 100% |  33.4 MiB/s |   1.4 MiB |  00m00s
[211/215] Installing gcc-c++-0:15.0.1-0 100% | 283.8 MiB/s |  41.1 MiB |  00m00s
[212/215] Installing annobin-plugin-gcc 100% |  36.0 MiB/s | 995.0 KiB |  00m00s
[213/215] Installing gcc-plugin-annobin 100% |   2.1 MiB/s |  58.8 KiB |  00m00s
[214/215] Installing langpacks-en-0:4.2 100% | 683.6 KiB/s | 700.0   B |  00m00s
[215/215] Installing xxd-2:9.1.1227-1.f 100% | 118.4 KiB/s |  34.3 KiB |  00m00s
Warning: skipped OpenPGP checks for 13 packages from repository: copr_base
Complete!
Finish: build setup for llama-cpp-b4580-2.fc43.src.rpm
Start: rpmbuild llama-cpp-b4580-2.fc43.src.rpm
Building target platforms: x86_64
Building for target x86_64
setting SOURCE_DATE_EPOCH=1741478400
Executing(%mkbuilddir): /bin/sh -e /var/tmp/rpm-tmp.uOUhdA
Executing(%prep): /bin/sh -e /var/tmp/rpm-tmp.Q3A1Sb
+ umask 022
+ cd /builddir/build/BUILD/llama-cpp-b4580-build
+ cd /builddir/build/BUILD/llama-cpp-b4580-build
+ rm -rf llama.cpp-b4580
+ /usr/lib/rpm/rpmuncompress -x /builddir/build/SOURCES/llama.cpp-b4580.tar.gz
+ STATUS=0
+ '[' 0 -ne 0 ']'
+ cd llama.cpp-b4580
+ /usr/bin/chmod -Rf a+rX,u+w,g-w,o-w .
+ sed -i -e 's/POSITION_INDEPENDENT_CODE ON/POSITION_INDEPENDENT_CODE ON SOVERSION b4580/' src/CMakeLists.txt
+ sed -i -e 's/POSITION_INDEPENDENT_CODE ON/POSITION_INDEPENDENT_CODE ON SOVERSION b4580/' ggml/src/CMakeLists.txt
+ sed -i '/target_link_libraries(ggml-hip PRIVATE ggml-base.*/aset_target_properties(ggml-hip PROPERTIES SOVERSION b4580)' ggml/src/ggml-hip/CMakeLists.txt
+ sed -i '/target_compile_features(${GGML_CPU_NAME} PRIVATE c_std_11.*/aset_target_properties(${GGML_CPU_NAME} PROPERTIES SOVERSION b4580)' ggml/src/ggml-cpu/CMakeLists.txt
+ sed -i '/#include <vector.*/a#include <cstdint>' src/llama-mmap.h
+ rm -rf exmples/llma.android
+ find . -name .gitignore -exec rm -rf '{}' ';'
+ RPM_EC=0
++ jobs -p
+ exit 0
Executing(%build): /bin/sh -e /var/tmp/rpm-tmp.fgtbJE
+ umask 022
+ cd /builddir/build/BUILD/llama-cpp-b4580-build
+ CFLAGS='-O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -fstack-protector-strong   -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer '
+ export CFLAGS
+ CXXFLAGS='-O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer'
+ export CXXFLAGS
+ FFLAGS='-O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -fstack-protector-strong   -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -I/usr/lib64/gfortran/modules '
+ export FFLAGS
+ FCFLAGS='-O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -fstack-protector-strong   -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -I/usr/lib64/gfortran/modules '
+ export FCFLAGS
+ VALAFLAGS=-g
+ export VALAFLAGS
+ RUSTFLAGS='-Copt-level=3 -Cdebuginfo=2 -Ccodegen-units=1 -Cstrip=none -Cforce-frame-pointers=yes -Clink-arg=-specs=/usr/lib/rpm/redhat/redhat-package-notes --cap-lints=warn'
+ export RUSTFLAGS
+ LDFLAGS='-Wl,-z,relro -Wl,--as-needed  -Wl,-z,pack-relative-relocs -Wl,-z,now -Wl,-z,now   -Wl,--build-id=sha1 -specs=/usr/lib/rpm/redhat/redhat-package-notes '
+ export LDFLAGS
+ LT_SYS_LIBRARY_PATH=/usr/lib64:
+ export LT_SYS_LIBRARY_PATH
+ CC=hipcc
+ export CC
+ CXX=hipcc
+ export CXX
+ cd llama.cpp-b4580
+ CFLAGS='-O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -fstack-protector-strong   -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer '
+ export CFLAGS
+ CXXFLAGS='-O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer'
+ export CXXFLAGS
+ FFLAGS='-O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -fstack-protector-strong   -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -I/usr/lib64/gfortran/modules '
+ export FFLAGS
+ FCFLAGS='-O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -fstack-protector-strong   -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -I/usr/lib64/gfortran/modules '
+ export FCFLAGS
+ VALAFLAGS=-g
+ export VALAFLAGS
+ RUSTFLAGS='-Copt-level=3 -Cdebuginfo=2 -Ccodegen-units=1 -Cstrip=none -Cforce-frame-pointers=yes -Clink-arg=-specs=/usr/lib/rpm/redhat/redhat-package-notes --cap-lints=warn'
+ export RUSTFLAGS
+ LDFLAGS='-Wl,-z,relro -Wl,--as-needed  -Wl,-z,pack-relative-relocs -Wl,-z,now -Wl,-z,now   -Wl,--build-id=sha1 -specs=/usr/lib/rpm/redhat/redhat-package-notes '
+ export LDFLAGS
+ LT_SYS_LIBRARY_PATH=/usr/lib64:
+ export LT_SYS_LIBRARY_PATH
+ CC=hipcc
+ export CC
+ CXX=hipcc
+ export CXX
+ /usr/bin/cmake -S . -B redhat-linux-build -DCMAKE_C_FLAGS_RELEASE:STRING=-DNDEBUG -DCMAKE_CXX_FLAGS_RELEASE:STRING=-DNDEBUG -DCMAKE_Fortran_FLAGS_RELEASE:STRING=-DNDEBUG -DCMAKE_VERBOSE_MAKEFILE:BOOL=ON -DCMAKE_INSTALL_DO_STRIP:BOOL=OFF -DCMAKE_INSTALL_PREFIX:PATH=/usr -DCMAKE_INSTALL_FULL_SBINDIR:PATH=/usr/bin -DCMAKE_INSTALL_SBINDIR:PATH=bin -DBUILD_SHARED_LIBS:BOOL=ON -DCMAKE_INSTALL_LIBDIR=lib64 -DCMAKE_SKIP_RPATH=ON -DGGML_AVX=OFF -DGGML_AVX2=OFF -DGGML_AVX512=OFF -DGGML_AVX512_VBMI=OFF -DGGML_AVX512_VNNI=OFF -DGGML_FMA=OFF -DGGML_F16C=OFF -DGGML_HIP=ON '-DAMDGPU_TARGETS=gfx900;gfx906:xnack-;gfx908:xnack-;gfx90a:xnack+;gfx90a:xnack-;gfx942;gfx1010;gfx1012;gfx1030;gfx1031;gfx1035;gfx1100;gfx1101;gfx1102;gfx1103;gfx1150;gfx1151;gfx1152;gfx1200;gfx1201' -DLLAMA_BUILD_EXAMPLES=OFF -DLLAMA_BUILD_TESTS=OFF
-- The C compiler identification is Clang 18.0.0
-- The CXX compiler identification is Clang 18.0.0
-- Detecting C compiler ABI info
-- Detecting C compiler ABI info - done
-- Check for working C compiler: /usr/bin/hipcc - skipped
-- Detecting C compile features
-- Detecting C compile features - done
-- Detecting CXX compiler ABI info
-- Detecting CXX compiler ABI info - done
-- Check for working CXX compiler: /usr/bin/hipcc - skipped
-- Detecting CXX compile features
-- Detecting CXX compile features - done
-- Found Git: /usr/bin/git (found version "2.49.0")
fatal: not a git repository (or any of the parent directories): .git
fatal: not a git repository (or any of the parent directories): .git
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
-- Setting GGML_NATIVE_DEFAULT to OFF
-- Performing Test CMAKE_HAVE_LIBC_PTHREAD
-- Performing Test CMAKE_HAVE_LIBC_PTHREAD - Success
-- Found Threads: TRUE
-- Warning: ccache not found - consider installing it for faster compilation or disable this warning with GGML_CCACHE=OFF
-- CMAKE_SYSTEM_PROCESSOR: x86_64
-- Including CPU backend
-- Could NOT find OpenMP_C (missing: OpenMP_C_FLAGS OpenMP_C_LIB_NAMES) 
-- Could NOT find OpenMP_CXX (missing: OpenMP_CXX_FLAGS OpenMP_CXX_LIB_NAMES) 
-- Could NOT find OpenMP (missing: OpenMP_C_FOUND OpenMP_CXX_FOUND) 
CMake Warning at ggml/src/ggml-cpu/CMakeLists.txt:54 (message):
  OpenMP not found
Call Stack (most recent call first):
  ggml/src/CMakeLists.txt:312 (ggml_add_cpu_backend_variant_impl)


-- x86 detected
-- Adding CPU backend variant ggml-cpu: -msse4.2 GGML_SSE42
CMake Warning at ggml/src/ggml-hip/CMakeLists.txt:27 (message):
  Setting hipcc as the C++ compiler is legacy behavior.  Prefer setting the
  HIP compiler directly.  See README for details.


-- Performing Test HIP_CLANG_SUPPORTS_PARALLEL_JOBS
-- Performing Test HIP_CLANG_SUPPORTS_PARALLEL_JOBS - Success
-- HIP and hipBLAS found
-- Including HIP backend
fatal: not a git repository (or any of the parent directories): .git
fatal: not a git repository (or any of the parent directories): .git
CMake Warning at common/CMakeLists.txt:32 (message):
  Git repository not found; to enable automatic generation of build info,
  make sure Git is installed and the project is a Git repository.


-- Configuring done (8.4s)
-- Generating done (0.0s)
CMake Warning:
  Manually-specified variables were not used by the project:

    CMAKE_Fortran_FLAGS_RELEASE
    CMAKE_INSTALL_DO_STRIP


-- Build files have been written to: /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build
+ /usr/bin/cmake --build redhat-linux-build -j2 --verbose
Change Dir: '/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build'

Run Build Command(s): /usr/bin/cmake -E env VERBOSE=1 /usr/bin/gmake -f Makefile -j2
/usr/bin/cmake -S/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580 -B/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build --check-build-system CMakeFiles/Makefile.cmake 0
/usr/bin/cmake -E cmake_progress_start /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/CMakeFiles /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build//CMakeFiles/progress.marks
/usr/bin/gmake  -f CMakeFiles/Makefile2 all
gmake[1]: Entering directory '/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build'
/usr/bin/gmake  -f ggml/src/CMakeFiles/ggml-base.dir/build.make ggml/src/CMakeFiles/ggml-base.dir/depend
/usr/bin/gmake  -f common/CMakeFiles/build_info.dir/build.make common/CMakeFiles/build_info.dir/depend
gmake[2]: Entering directory '/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build'
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build && /usr/bin/cmake -E cmake_depends "Unix Makefiles" /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580 /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/CMakeFiles/ggml-base.dir/DependInfo.cmake "--color="
gmake[2]: Entering directory '/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build'
gmake[2]: Leaving directory '/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build'
/usr/bin/gmake  -f ggml/src/CMakeFiles/ggml-base.dir/build.make ggml/src/CMakeFiles/ggml-base.dir/build
gmake[2]: Entering directory '/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build'
[  0%] Generating build details from Git
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580 && /usr/bin/cmake -DMSVC= -DCMAKE_C_COMPILER_VERSION=18.0.0 -DCMAKE_C_COMPILER_ID=Clang -DCMAKE_VS_PLATFORM_NAME= -DCMAKE_C_COMPILER=/usr/bin/hipcc -P /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/common/cmake/build-info-gen-cpp.cmake
[  1%] Building C object ggml/src/CMakeFiles/ggml-base.dir/ggml.c.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src && /usr/bin/hipcc -DGGML_BUILD -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_base_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -fstack-protector-strong   -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu11 -fPIC -Wshadow -Wstrict-prototypes -Wpointer-arith -Wmissing-prototypes -Werror=implicit-int -Werror=implicit-function-declaration -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wdouble-promotion -MD -MT ggml/src/CMakeFiles/ggml-base.dir/ggml.c.o -MF CMakeFiles/ggml-base.dir/ggml.c.o.d -o CMakeFiles/ggml-base.dir/ggml.c.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml.c
-- Found Git: /usr/bin/git (found version "2.49.0")
fatal: not a git repository (or any of the parent directories): .git
fatal: not a git repository (or any of the parent directories): .git
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build && /usr/bin/cmake -E cmake_depends "Unix Makefiles" /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580 /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/common /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/common /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/common/CMakeFiles/build_info.dir/DependInfo.cmake "--color="
gmake[2]: Leaving directory '/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build'
/usr/bin/gmake  -f common/CMakeFiles/build_info.dir/build.make common/CMakeFiles/build_info.dir/build
gmake[2]: Entering directory '/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build'
[  2%] Building CXX object common/CMakeFiles/build_info.dir/build-info.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/common && /usr/bin/hipcc   -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT common/CMakeFiles/build_info.dir/build-info.cpp.o -MF CMakeFiles/build_info.dir/build-info.cpp.o.d -o CMakeFiles/build_info.dir/build-info.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/common/build-info.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[  3%] Building C object ggml/src/CMakeFiles/ggml-base.dir/ggml-alloc.c.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src && /usr/bin/hipcc -DGGML_BUILD -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_base_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -fstack-protector-strong   -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu11 -fPIC -Wshadow -Wstrict-prototypes -Wpointer-arith -Wmissing-prototypes -Werror=implicit-int -Werror=implicit-function-declaration -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wdouble-promotion -MD -MT ggml/src/CMakeFiles/ggml-base.dir/ggml-alloc.c.o -MF CMakeFiles/ggml-base.dir/ggml-alloc.c.o.d -o CMakeFiles/ggml-base.dir/ggml-alloc.c.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-alloc.c
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[  4%] Building CXX object ggml/src/CMakeFiles/ggml-base.dir/ggml-backend.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src && /usr/bin/hipcc -DGGML_BUILD -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_base_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT ggml/src/CMakeFiles/ggml-base.dir/ggml-backend.cpp.o -MF CMakeFiles/ggml-base.dir/ggml-backend.cpp.o.d -o CMakeFiles/ggml-base.dir/ggml-backend.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-backend.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
gmake[2]: Leaving directory '/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build'
[  4%] Built target build_info
[  4%] Building CXX object ggml/src/CMakeFiles/ggml-base.dir/ggml-opt.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src && /usr/bin/hipcc -DGGML_BUILD -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_base_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT ggml/src/CMakeFiles/ggml-base.dir/ggml-opt.cpp.o -MF CMakeFiles/ggml-base.dir/ggml-opt.cpp.o.d -o CMakeFiles/ggml-base.dir/ggml-opt.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-opt.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[  5%] Building CXX object ggml/src/CMakeFiles/ggml-base.dir/ggml-threading.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src && /usr/bin/hipcc -DGGML_BUILD -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_base_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT ggml/src/CMakeFiles/ggml-base.dir/ggml-threading.cpp.o -MF CMakeFiles/ggml-base.dir/ggml-threading.cpp.o.d -o CMakeFiles/ggml-base.dir/ggml-threading.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-threading.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[  6%] Building C object ggml/src/CMakeFiles/ggml-base.dir/ggml-quants.c.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src && /usr/bin/hipcc -DGGML_BUILD -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_base_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -fstack-protector-strong   -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu11 -fPIC -Wshadow -Wstrict-prototypes -Wpointer-arith -Wmissing-prototypes -Werror=implicit-int -Werror=implicit-function-declaration -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wdouble-promotion -MD -MT ggml/src/CMakeFiles/ggml-base.dir/ggml-quants.c.o -MF CMakeFiles/ggml-base.dir/ggml-quants.c.o.d -o CMakeFiles/ggml-base.dir/ggml-quants.c.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-quants.c
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[  7%] Building CXX object ggml/src/CMakeFiles/ggml-base.dir/gguf.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src && /usr/bin/hipcc -DGGML_BUILD -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_base_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT ggml/src/CMakeFiles/ggml-base.dir/gguf.cpp.o -MF CMakeFiles/ggml-base.dir/gguf.cpp.o.d -o CMakeFiles/ggml-base.dir/gguf.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/gguf.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[  8%] Linking CXX shared library ../../bin/libggml-base.so
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src && /usr/bin/cmake -E cmake_link_script CMakeFiles/ggml-base.dir/link.txt --verbose=1
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
clang++: warning: argument unused during compilation: '-Xarch_host -fstack-protector-strong' [-Wunused-command-line-argument]
clang++: warning: argument unused during compilation: '-Xarch_host -fcf-protection' [-Wunused-command-line-argument]
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]
/usr/bin/hipcc -fPIC -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -Xlinker --dependency-file=CMakeFiles/ggml-base.dir/link.d -Wl,-z,relro -Wl,--as-needed  -Wl,-z,pack-relative-relocs -Wl,-z,now -Wl,-z,now   -Wl,--build-id=sha1 -specs=/usr/lib/rpm/redhat/redhat-package-notes -shared -Wl,-soname,libggml-base.so.b4580 -o ../../bin/libggml-base.so.b4580 "CMakeFiles/ggml-base.dir/ggml.c.o" "CMakeFiles/ggml-base.dir/ggml-alloc.c.o" "CMakeFiles/ggml-base.dir/ggml-backend.cpp.o" "CMakeFiles/ggml-base.dir/ggml-opt.cpp.o" "CMakeFiles/ggml-base.dir/ggml-threading.cpp.o" "CMakeFiles/ggml-base.dir/ggml-quants.c.o" "CMakeFiles/ggml-base.dir/gguf.cpp.o"  -lm
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src && /usr/bin/cmake -E cmake_symlink_library ../../bin/libggml-base.so.b4580 ../../bin/libggml-base.so.b4580 ../../bin/libggml-base.so
gmake[2]: Leaving directory '/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build'
[  8%] Built target ggml-base
/usr/bin/gmake  -f ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/build.make ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/depend
/usr/bin/gmake  -f ggml/src/CMakeFiles/ggml-cpu.dir/build.make ggml/src/CMakeFiles/ggml-cpu.dir/depend
gmake[2]: Entering directory '/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build'
gmake[2]: Entering directory '/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build'
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build && /usr/bin/cmake -E cmake_depends "Unix Makefiles" /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580 /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/DependInfo.cmake "--color="
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build && /usr/bin/cmake -E cmake_depends "Unix Makefiles" /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580 /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/CMakeFiles/ggml-cpu.dir/DependInfo.cmake "--color="
gmake[2]: Leaving directory '/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build'
/usr/bin/gmake  -f ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/build.make ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/build
gmake[2]: Entering directory '/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build'
gmake[2]: Leaving directory '/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build'
/usr/bin/gmake  -f ggml/src/CMakeFiles/ggml-cpu.dir/build.make ggml/src/CMakeFiles/ggml-cpu.dir/build
gmake[2]: Entering directory '/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build'
[  9%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/acc.cu.o
[ 10%] Building C object ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu.c.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/acc.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/acc.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/acc.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cu
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_SSE42 -DGGML_USE_CPU_AARCH64 -DGGML_USE_LLAMAFILE -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_cpu_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cpu -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -fstack-protector-strong   -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu11 -fPIC -Wshadow -Wstrict-prototypes -Wpointer-arith -Wmissing-prototypes -Werror=implicit-int -Werror=implicit-function-declaration -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wdouble-promotion -msse4.2 -MD -MT ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu.c.o -MF CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu.c.o.d -o CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu.c.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cpu/ggml-cpu.c
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1012.
[ 10%] Building CXX object ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_SSE42 -DGGML_USE_CPU_AARCH64 -DGGML_USE_LLAMAFILE -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_cpu_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cpu -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -msse4.2 -MD -MT ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu.cpp.o -MF CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu.cpp.o.d -o CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cpu/ggml-cpu.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1031.
[ 11%] Building CXX object ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu-aarch64.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_SSE42 -DGGML_USE_CPU_AARCH64 -DGGML_USE_LLAMAFILE -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_cpu_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cpu -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -msse4.2 -MD -MT ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu-aarch64.cpp.o -MF CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu-aarch64.cpp.o.d -o CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu-aarch64.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cpu/ggml-cpu-aarch64.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1100.
[ 12%] Building CXX object ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu-hbm.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_SSE42 -DGGML_USE_CPU_AARCH64 -DGGML_USE_LLAMAFILE -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_cpu_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cpu -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -msse4.2 -MD -MT ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu-hbm.cpp.o -MF CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu-hbm.cpp.o.d -o CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu-hbm.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cpu/ggml-cpu-hbm.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1101.
[ 13%] Building C object ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu-quants.c.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_SSE42 -DGGML_USE_CPU_AARCH64 -DGGML_USE_LLAMAFILE -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_cpu_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cpu -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -fstack-protector-strong   -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu11 -fPIC -Wshadow -Wstrict-prototypes -Wpointer-arith -Wmissing-prototypes -Werror=implicit-int -Werror=implicit-function-declaration -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wdouble-promotion -msse4.2 -MD -MT ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu-quants.c.o -MF CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu-quants.c.o.d -o CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu-quants.c.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cpu/ggml-cpu-quants.c
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
[ 14%] Building CXX object ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu-traits.cpp.o
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_SSE42 -DGGML_USE_CPU_AARCH64 -DGGML_USE_LLAMAFILE -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_cpu_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cpu -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -msse4.2 -MD -MT ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu-traits.cpp.o -MF CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu-traits.cpp.o.d -o CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu-traits.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cpu/ggml-cpu-traits.cpp
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
6 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
[ 14%] Building CXX object ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/amx/amx.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_SSE42 -DGGML_USE_CPU_AARCH64 -DGGML_USE_LLAMAFILE -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_cpu_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cpu -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -msse4.2 -MD -MT ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/amx/amx.cpp.o -MF CMakeFiles/ggml-cpu.dir/ggml-cpu/amx/amx.cpp.o.d -o CMakeFiles/ggml-cpu.dir/ggml-cpu/amx/amx.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cpu/amx/amx.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1151.
[ 15%] Building CXX object ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/amx/mmq.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_SSE42 -DGGML_USE_CPU_AARCH64 -DGGML_USE_LLAMAFILE -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_cpu_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cpu -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -msse4.2 -MD -MT ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/amx/mmq.cpp.o -MF CMakeFiles/ggml-cpu.dir/ggml-cpu/amx/mmq.cpp.o.d -o CMakeFiles/ggml-cpu.dir/ggml-cpu/amx/mmq.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cpu/amx/mmq.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1200.
[ 16%] Building CXX object ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/llamafile/sgemm.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_SSE42 -DGGML_USE_CPU_AARCH64 -DGGML_USE_LLAMAFILE -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_cpu_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cpu -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -msse4.2 -MD -MT ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/llamafile/sgemm.cpp.o -MF CMakeFiles/ggml-cpu.dir/ggml-cpu/llamafile/sgemm.cpp.o.d -o CMakeFiles/ggml-cpu.dir/ggml-cpu/llamafile/sgemm.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cpu/llamafile/sgemm.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
[ 17%] Linking CXX shared library ../../bin/libggml-cpu.so
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src && /usr/bin/cmake -E cmake_link_script CMakeFiles/ggml-cpu.dir/link.txt --verbose=1
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
clang++: warning: argument unused during compilation: '-Xarch_host -fstack-protector-strong' [-Wunused-command-line-argument]
clang++: warning: argument unused during compilation: '-Xarch_host -fcf-protection' [-Wunused-command-line-argument]
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]
6 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
/usr/bin/hipcc -fPIC -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -Xlinker --dependency-file=CMakeFiles/ggml-cpu.dir/link.d -Wl,-z,relro -Wl,--as-needed  -Wl,-z,pack-relative-relocs -Wl,-z,now -Wl,-z,now   -Wl,--build-id=sha1 -specs=/usr/lib/rpm/redhat/redhat-package-notes -shared -Wl,-soname,libggml-cpu.so.b4580 -o ../../bin/libggml-cpu.so.b4580 "CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu.c.o" "CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu.cpp.o" "CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu-aarch64.cpp.o" "CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu-hbm.cpp.o" "CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu-quants.c.o" "CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu-traits.cpp.o" "CMakeFiles/ggml-cpu.dir/ggml-cpu/amx/amx.cpp.o" "CMakeFiles/ggml-cpu.dir/ggml-cpu/amx/mmq.cpp.o" "CMakeFiles/ggml-cpu.dir/ggml-cpu/llamafile/sgemm.cpp.o"  ../../bin/libggml-base.so.b4580
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src && /usr/bin/cmake -E cmake_symlink_library ../../bin/libggml-cpu.so.b4580 ../../bin/libggml-cpu.so.b4580 ../../bin/libggml-cpu.so
gmake[2]: Leaving directory '/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build'
[ 17%] Built target ggml-cpu
[ 18%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/arange.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/arange.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/arange.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/arange.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx942.
6 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/acc.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 18%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/argmax.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/argmax.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/argmax.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/argmax.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cu
6 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1010.
6 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1012.
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1030.
6 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1031.
6 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1035.
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1100.
6 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1150.
6 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1151.
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1152.
6 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1200.
6 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1201.
6 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx900.
6 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
6 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
6 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
6 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx942.
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/arange.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 19%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/argsort.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/argsort.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/argsort.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/argsort.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cu
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1010.
6 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1012.
6 warnings generated when compiling for host.
[ 20%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/binbcast.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/binbcast.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/binbcast.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/binbcast.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1030.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:359:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  359 |         } break;
      |           ^~~~~
7 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:359:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  359 |         } break;
      |           ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1035.
7 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:359:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  359 |         } break;
      |           ^~~~~
6 warnings generated when compiling for gfx1100.
7 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:359:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  359 |         } break;
      |           ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
7 warnings generated when compiling for gfx1031.
6 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:359:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  359 |         } break;
      |           ^~~~~
6 warnings generated when compiling for gfx1103.
7 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1150.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:359:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  359 |         } break;
      |           ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
7 warnings generated when compiling for gfx1100.
6 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:359:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  359 |         } break;
      |           ^~~~~
6 warnings generated when compiling for gfx1152.
7 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1200.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:359:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  359 |         } break;
      |           ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
7 warnings generated when compiling for gfx1102.
6 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:359:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  359 |         } break;
      |           ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx900.
7 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:359:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  359 |         } break;
      |           ^~~~~
7 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:359:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  359 |         } break;
      |           ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
7 warnings generated when compiling for gfx1151.
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:359:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  359 |         } break;
      |           ^~~~~
6 warnings generated when compiling for gfx90a.
7 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
6 warnings generated when compiling for gfx942.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:359:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  359 |         } break;
      |           ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
7 warnings generated when compiling for gfx1200.
6 warnings generated when compiling for host.
[ 21%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/clamp.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/clamp.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/clamp.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/clamp.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:359:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  359 |         } break;
      |           ^~~~~
6 warnings generated when compiling for gfx1010.
7 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1012.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:359:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  359 |         } break;
      |           ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
7 warnings generated when compiling for gfx900.
6 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:359:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  359 |         } break;
      |           ^~~~~
6 warnings generated when compiling for gfx1031.
7 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1035.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:359:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  359 |         } break;
      |           ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
7 warnings generated when compiling for gfx908.
6 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:359:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  359 |         } break;
      |           ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1101.
7 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1102.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:359:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  359 |         } break;
      |           ^~~~~
7 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:359:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  359 |         } break;
      |           ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1150.
7 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1151.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/binbcast.cu:359:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  359 |         } break;
      |           ^~~~~
7 warnings generated when compiling for host.
[ 22%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/concat.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/concat.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/concat.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/concat.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
9 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
9 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
9 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
9 warnings generated when compiling for gfx1031.
6 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
9 warnings generated when compiling for gfx1035.
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
9 warnings generated when compiling for gfx1100.
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
6 warnings generated when compiling for gfx90a.
9 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
6 warnings generated when compiling for gfx942.
9 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/clamp.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 22%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/conv-transpose-1d.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/conv-transpose-1d.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/conv-transpose-1d.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/conv-transpose-1d.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
9 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for gfx1010.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
9 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for gfx1012.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
9 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
17 warnings generated when compiling for gfx1030.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
9 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
9 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
9 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
9 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
9 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
9 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
9 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
9 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
9 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:41:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   41 |     if (blockIdx.y < ne01) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:67:20: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   67 |     if (blockIdx.z < ne02) { // src0
      |         ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
9 warnings generated when compiling for host.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
[ 23%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/convert.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/convert.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/convert.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/convert.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'hip_bfloat16 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<hip_bfloat16, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:684:20: note: in instantiation of function template specialization 'convert_unary_cuda<hip_bfloat16, float>' requested here
  684 |             return convert_unary_cuda<nv_bfloat16>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for gfx1201.
13 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for gfx900.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'hip_bfloat16 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<hip_bfloat16, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:684:20: note: in instantiation of function template specialization 'convert_unary_cuda<hip_bfloat16, float>' requested here
  684 |             return convert_unary_cuda<nv_bfloat16>;
      |                    ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
13 warnings generated when compiling for gfx1012.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'hip_bfloat16 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<hip_bfloat16, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:684:20: note: in instantiation of function template specialization 'convert_unary_cuda<hip_bfloat16, float>' requested here
  684 |             return convert_unary_cuda<nv_bfloat16>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for gfx908.
13 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'hip_bfloat16 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<hip_bfloat16, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:684:20: note: in instantiation of function template specialization 'convert_unary_cuda<hip_bfloat16, float>' requested here
  684 |             return convert_unary_cuda<nv_bfloat16>;
      |                    ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
13 warnings generated when compiling for gfx1031.
17 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'hip_bfloat16 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<hip_bfloat16, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:684:20: note: in instantiation of function template specialization 'convert_unary_cuda<hip_bfloat16, float>' requested here
  684 |             return convert_unary_cuda<nv_bfloat16>;
      |                    ^
17 warnings generated when compiling for gfx942.
13 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:33: warning: unused parameter 'p0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:4:47: warning: unused parameter 'd0' [-Wunused-parameter]
    4 |         const int s0, const int p0, const int d0, const int output_size,
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:5:79: warning: unused parameter 'src0_ne3' [-Wunused-parameter]
    5 |         const int src0_ne0, const int src0_ne1, const int src0_ne2, const int src0_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:39: warning: unused parameter 'src1_ne1' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:59: warning: unused parameter 'src1_ne2' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:6:79: warning: unused parameter 'src1_ne3' [-Wunused-parameter]
    6 |         const int src1_ne0, const int src1_ne1, const int src1_ne2, const int src1_ne3,
      |                                                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:38: warning: unused parameter 'dst_ne1' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                      ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:57: warning: unused parameter 'dst_ne2' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:7:76: warning: unused parameter 'dst_ne3' [-Wunused-parameter]
    7 |         const int dst_ne0, const int dst_ne1, const int dst_ne2, const int dst_ne3,
      |                                                                            ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:78:19: warning: unused variable 'kernel_size' [-Wunused-variable]
   78 |     const int64_t kernel_size = ggml_nelements(src0);
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/conv-transpose-1d.cu:79:19: warning: unused variable 'input_size' [-Wunused-variable]
   79 |     const int64_t input_size = ggml_nelements(src1);
      |                   ^~~~~~~~~~
17 warnings generated when compiling for host.
[ 24%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/count-equal.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/count-equal.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/count-equal.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/count-equal.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'hip_bfloat16 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<hip_bfloat16, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:684:20: note: in instantiation of function template specialization 'convert_unary_cuda<hip_bfloat16, float>' requested here
  684 |             return convert_unary_cuda<nv_bfloat16>;
      |                    ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
13 warnings generated when compiling for gfx1100.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'hip_bfloat16 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<hip_bfloat16, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:684:20: note: in instantiation of function template specialization 'convert_unary_cuda<hip_bfloat16, float>' requested here
  684 |             return convert_unary_cuda<nv_bfloat16>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx1012.
13 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
7 warnings generated when compiling for gfx1030.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'hip_bfloat16 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<hip_bfloat16, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:684:20: note: in instantiation of function template specialization 'convert_unary_cuda<hip_bfloat16, float>' requested here
  684 |             return convert_unary_cuda<nv_bfloat16>;
      |                    ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
13 warnings generated when compiling for gfx1102.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'hip_bfloat16 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<hip_bfloat16, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:684:20: note: in instantiation of function template specialization 'convert_unary_cuda<hip_bfloat16, float>' requested here
  684 |             return convert_unary_cuda<nv_bfloat16>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx1035.
13 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
7 warnings generated when compiling for gfx1100.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'hip_bfloat16 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<hip_bfloat16, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:684:20: note: in instantiation of function template specialization 'convert_unary_cuda<hip_bfloat16, float>' requested here
  684 |             return convert_unary_cuda<nv_bfloat16>;
      |                    ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
13 warnings generated when compiling for gfx1150.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'hip_bfloat16 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<hip_bfloat16, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:684:20: note: in instantiation of function template specialization 'convert_unary_cuda<hip_bfloat16, float>' requested here
  684 |             return convert_unary_cuda<nv_bfloat16>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx1102.
13 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx1103.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'hip_bfloat16 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<hip_bfloat16, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:684:20: note: in instantiation of function template specialization 'convert_unary_cuda<hip_bfloat16, float>' requested here
  684 |             return convert_unary_cuda<nv_bfloat16>;
      |                    ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
13 warnings generated when compiling for gfx1152.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'hip_bfloat16 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<hip_bfloat16, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:684:20: note: in instantiation of function template specialization 'convert_unary_cuda<hip_bfloat16, float>' requested here
  684 |             return convert_unary_cuda<nv_bfloat16>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx1151.
13 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx1152.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'hip_bfloat16 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<hip_bfloat16, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:684:20: note: in instantiation of function template specialization 'convert_unary_cuda<hip_bfloat16, float>' requested here
  684 |             return convert_unary_cuda<nv_bfloat16>;
      |                    ^
13 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'hip_bfloat16 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<hip_bfloat16, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:684:20: note: in instantiation of function template specialization 'convert_unary_cuda<hip_bfloat16, float>' requested here
  684 |             return convert_unary_cuda<nv_bfloat16>;
      |                    ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx1201.
13 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
7 warnings generated when compiling for gfx900.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'hip_bfloat16 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<hip_bfloat16, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:684:20: note: in instantiation of function template specialization 'convert_unary_cuda<hip_bfloat16, float>' requested here
  684 |             return convert_unary_cuda<nv_bfloat16>;
      |                    ^
13 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'hip_bfloat16 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<hip_bfloat16, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:684:20: note: in instantiation of function template specialization 'convert_unary_cuda<hip_bfloat16, float>' requested here
  684 |             return convert_unary_cuda<nv_bfloat16>;
      |                    ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
13 warnings generated when compiling for gfx908.
7 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
7 warnings generated when compiling for gfx90a.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'hip_bfloat16 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<hip_bfloat16, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:684:20: note: in instantiation of function template specialization 'convert_unary_cuda<hip_bfloat16, float>' requested here
  684 |             return convert_unary_cuda<nv_bfloat16>;
      |                    ^
13 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'hip_bfloat16 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<hip_bfloat16, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:684:20: note: in instantiation of function template specialization 'convert_unary_cuda<hip_bfloat16, float>' requested here
  684 |             return convert_unary_cuda<nv_bfloat16>;
      |                    ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
13 warnings generated when compiling for gfx90a.
7 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'hip_bfloat16 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<hip_bfloat16, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:684:20: note: in instantiation of function template specialization 'convert_unary_cuda<hip_bfloat16, float>' requested here
  684 |             return convert_unary_cuda<nv_bfloat16>;
      |                    ^
7 warnings generated when compiling for host.
[ 25%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/cpy.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/cpy.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/cpy.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/cpy.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cu
13 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *__restrict' to 'type-parameter-0-0 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
6 warnings generated when compiling for gfx1010.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:467:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  467 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:33:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   33 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:470:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  470 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'float *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<float, __half>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:635:20: note: in instantiation of function template specialization 'convert_unary_cuda<float, __half>' requested here
  635 |             return convert_unary_cuda<float>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to '__half *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<__half, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:682:20: note: in instantiation of function template specialization 'convert_unary_cuda<__half, float>' requested here
  682 |             return convert_unary_cuda<half>;
      |                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:580:33: warning: cast from 'const void *' to 'hip_bfloat16 *' drops const qualifier [-Wcast-qual]
  580 |     const src_t * x = (src_t *) vx;
      |                                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:588:5: note: in instantiation of function template specialization 'convert_unary<hip_bfloat16, float>' requested here
  588 |     convert_unary<src_t><<<num_blocks, CUDA_DEQUANTIZE_BLOCK_SIZE, 0, stream>>>(vx, y, k);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/convert.cu:684:20: note: in instantiation of function template specialization 'convert_unary_cuda<hip_bfloat16, float>' requested here
  684 |             return convert_unary_cuda<nv_bfloat16>;
      |                    ^
13 warnings generated when compiling for host.
[ 26%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/cross-entropy-loss.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/cross-entropy-loss.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/cross-entropy-loss.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/cross-entropy-loss.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cross-entropy-loss.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1010.
6 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1012.
6 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1030.
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1035.
6 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1100.
6 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1101.
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1102.
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
6 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1151.
6 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1152.
6 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1152.
6 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx900.
6 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
6 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx900.
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx942.
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cross-entropy-loss.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 27%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/diagmask.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/diagmask.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/diagmask.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/diagmask.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cu
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
6 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1030.
6 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cpy.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1031.
6 warnings generated when compiling for host.
[ 27%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn-tile-f16.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn-tile-f16.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn-tile-f16.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn-tile-f16.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1102.
30 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1152.
30 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
30 warnings generated when compiling for gfx1030.
6 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
30 warnings generated when compiling for gfx1031.
6 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
6 warnings generated when compiling for host.
[ 28%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn-tile-f32.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn-tile-f32.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn-tile-f32.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn-tile-f32.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
30 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
30 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
30 warnings generated when compiling for gfx1030.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
30 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
30 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
30 warnings generated when compiling for gfx1100.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
30 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
30 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
30 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
30 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx900.
30 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx906.
30 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx1200.
30 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx1201.
30 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
30 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:323:13: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<16, 4, false>' requested here
  323 |             launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:294:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  294 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:300:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  300 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:348:9: note: in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
  348 |         launch_fattn_tile_f16_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
30 warnings generated when compiling for host.
[ 29%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
30 warnings generated when compiling for gfx908.
80 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_ca/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cus:e24(:c19o:n swarning: tunused parameter 'ne00' [-Wunused-parameter] 
int D   )24  | { 
        |                                                ^ 
   const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
80 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
80 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
30 warnings generated when compiling for gfx90a.
80 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
80 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
80 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
30 warnings generated when compiling for gfx90a.
80 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
80 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
80 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
30 warnings generated when compiling for gfx942.
80 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:24:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   24 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:27:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:28:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   28 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:31:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   31 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:32:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   32 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:33:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   33 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:36:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   36 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:39:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   39 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:40:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   40 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:41:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   41 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:42:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   42 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:43:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   43 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:44:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   44 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:45:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   45 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:46:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   46 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:319:13: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<16, 4, false>' requested here
  319 |             launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:293:13: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  293 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:299:13: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  299 |             launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f32.cu:344:9: note: in instantiation of function template specialization 'launch_fattn_tile_f32_64_128<32, 1, false>' requested here
  344 |         launch_fattn_tile_f32_64_128<cols_per_block, parallel_blocks, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
30 warnings generated when compiling for host.
[ 30%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/getrows.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/getrows.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/getrows.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/getrows.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu
80 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:201:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  201 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
80 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:201:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  201 |             break;
      |             ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
7 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
80 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:201:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  201 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
80 warnings generated when compiling for gfx1201.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:201:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  201 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:201:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  201 |             break;
      |             ^~~~~
80 warnings generated when compiling for gfx900.
7 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:201:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  201 |             break;
      |             ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
7 warnings generated when compiling for gfx1100.
80 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:201:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  201 |             break;
      |             ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
7 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
80 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:201:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  201 |             break;
      |             ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
7 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
80 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:201:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  201 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
80 warnings generated when compiling for gfx90a.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:201:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  201 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:201:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  201 |             break;
      |             ^~~~~
80 warnings generated when compiling for gfx942.
7 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:6:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:7:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:145:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  145 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:118:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  118 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:90:17: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:67:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:42:21: warning: 'break' will never be executed [-Wunreachable-code-break]
   42 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:201:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  201 |             break;
      |             ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn.cu:141:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_wmma_f16_case<256, 32, __half>' requested here
  141 |             ggml_cuda_flash_attn_ext_wmma_f16_case<256, cols_per_block, half>(ctx, dst);
      |             ^
80 warnings generated when compiling for host.
[ 31%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/ggml-cuda.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/ggml-cuda.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/ggml-cuda.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/ggml-cuda.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu
7 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:201:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  201 |             break;
      |             ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  132 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:131:9: note: declared here
  131 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:67: warning: unused parameter 'size' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
7 warnings generated when compiling for gfx1200.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3142:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3142 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3137:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3137 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3134:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3134 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3125:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3125 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3118:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3118 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3113:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3113 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3108:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3108 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3103:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3103 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3061:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3061 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3057:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3057 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3040:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3040 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2978:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2978 |             break;
      |             ^~~~~
29 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:201:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  201 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  132 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:131:9: note: declared here
  131 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:67: warning: unused parameter 'size' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3142:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3142 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3137:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3137 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3134:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3134 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3125:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3125 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3118:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3118 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3113:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3113 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3108:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3108 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3103:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3103 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3061:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3061 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3057:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3057 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3040:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3040 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2978:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2978 |             break;
      |             ^~~~~
29 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:201:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  201 |             break;
      |             ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
7 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  132 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:131:9: note: declared here
  131 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:67: warning: unused parameter 'size' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3142:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3142 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3137:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3137 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3134:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3134 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3125:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3125 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3118:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3118 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3113:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3113 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3108:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3108 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3103:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3103 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3061:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3061 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3057:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3057 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3040:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3040 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2978:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2978 |             break;
      |             ^~~~~
29 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:201:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  201 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  132 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:131:9: note: declared here
  131 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:67: warning: unused parameter 'size' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3142:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3142 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3137:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3137 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3134:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3134 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3125:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3125 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3118:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3118 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3113:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3113 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3108:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3108 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3103:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3103 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3061:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3061 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3057:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3057 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3040:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3040 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2978:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2978 |             break;
      |             ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
29 warnings generated when compiling for gfx1031.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:201:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  201 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  132 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:131:9: note: declared here
  131 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:67: warning: unused parameter 'size' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3142:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3142 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3137:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3137 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3134:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3134 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3125:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3125 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3118:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3118 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3113:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3113 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3108:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3108 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3103:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3103 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3061:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3061 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3057:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3057 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3040:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3040 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2978:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2978 |             break;
      |             ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
29 warnings generated when compiling for gfx1035.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:201:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  201 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
:17: warning: unused parameter 'ne00' [-Wunused-parameter]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9:  2497warning:  | anonymous types declared in an anonymous union are an extension [-Wnested-anon-types] 
   c  o298n | s t   i n t   &  snter0u0c,t  c{o
n      s| t        ^ 
int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  132 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:131:9: note: declared here
  131 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:67: warning: unused parameter 'size' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3142:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3142 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3137:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3137 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3134:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3134 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3125:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3125 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3118:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3118 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3113:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3113 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3108:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3108 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3103:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3103 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3061:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3061 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3057:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3057 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3040:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3040 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2978:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2978 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:201:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  201 |             break;
      |             ^~~~~
29 warnings generated when compiling for gfx1100.
7 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  132 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:131:9: note: declared here
  131 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:201:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  201 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:67: warning: unused parameter 'size' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3142:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3142 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3137:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3137 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3134:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3134 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3125:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3125 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3118:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3118 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3113:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3113 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3108:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3108 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3103:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3103 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3061:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3061 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3057:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3057 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3040:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3040 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2978:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2978 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx942.
29 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/getrows.cu:201:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  201 |             break;
      |             ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
7 warnings generated when compiling for host.
[ 31%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/gla.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/gla.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/gla.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/gla.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/gla.cu
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  132 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:131:9: note: declared here
  131 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:67: warning: unused parameter 'size' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3142:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3142 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3137:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3137 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3134:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3134 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3125:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3125 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3118:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3118 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3113:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3113 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3108:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3108 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3103:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3103 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3061:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3061 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3057:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3057 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3040:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3040 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2978:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2978 |             break;
      |             ^~~~~
29 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/gla.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  132 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:131:9: note: declared here
  131 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:67: warning: unused parameter 'size' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3142:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3142 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3137:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3137 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3134:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3134 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3125:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3125 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3118:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3118 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3113:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3113 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3108:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3108 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3103:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3103 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3061:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3061 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3057:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3057 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3040:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3040 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2978:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2978 |             break;
      |             ^~~~~
29 warnings generated when compiling for gfx1103.
6 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/gla.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  132 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:131:9: note: declared here
  131 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:67: warning: unused parameter 'size' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3142:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3142 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3137:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3137 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3134:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3134 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3125:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3125 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3118:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3118 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3113:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3113 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3108:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3108 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3103:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3103 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3061:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3061 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3057:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3057 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3040:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3040 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2978:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2978 |             break;
      |             ^~~~~
29 warnings generated when compiling for gfx1150.
6 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  132 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:131:9: note: declared here
  131 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:67: warning: unused parameter 'size' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3142:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3142 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3137:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3137 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3134:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3134 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3125:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3125 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3118:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3118 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3113:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3113 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3108:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3108 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3103:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3103 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3061:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3061 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3057:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3057 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3040:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3040 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2978:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2978 |             break;
      |             ^~~~~
29 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/gla.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  132 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:131:9: note: declared here
  131 |     int archLen = strlen(devName);
      |         ^
6 warnings generated when compiling for gfx1030.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:67: warning: unused parameter 'size' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3142:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3142 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3137:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3137 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3134:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3134 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3125:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3125 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3118:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3118 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3113:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3113 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3108:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3108 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3103:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3103 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3061:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3061 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3057:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3057 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3040:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3040 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2978:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2978 |             break;
      |             ^~~~~
29 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/gla.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  132 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:131:9: note: declared here
  131 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:67: warning: unused parameter 'size' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3142:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3142 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3137:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3137 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3134:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3134 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3125:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3125 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3118:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3118 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3113:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3113 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3108:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3108 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3103:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3103 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3061:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3061 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3057:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3057 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3040:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3040 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2978:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2978 |             break;
      |             ^~~~~
29 warnings generated when compiling for gfx1200.
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/gla.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  132 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:131:9: note: declared here
  131 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:67: warning: unused parameter 'size' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3142:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3142 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3137:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3137 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3134:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3134 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3125:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3125 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3118:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3118 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3113:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3113 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3108:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3108 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3103:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3103 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3061:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3061 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3057:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3057 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3040:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3040 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2978:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2978 |             break;
      |             ^~~~~
29 warnings generated when compiling for gfx1201.
6 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  132 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:131:9: note: declared here
  131 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:67: warning: unused parameter 'size' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3142:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3142 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3137:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3137 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3134:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3134 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3125:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3125 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3118:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3118 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3113:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3113 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3108:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3108 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3103:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3103 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3061:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3061 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3057:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3057 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3040:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3040 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2978:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2978 |             break;
      |             ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/gla.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
29 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  132 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:131:9: note: declared here
  131 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:67: warning: unused parameter 'size' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3142:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3142 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3137:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3137 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3134:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3134 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3125:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3125 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3118:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3118 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3113:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3113 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3108:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3108 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3103:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3103 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3061:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3061 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3057:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3057 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3040:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3040 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2978:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2978 |             break;
      |             ^~~~~
29 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/gla.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  132 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:131:9: note: declared here
  131 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:67: warning: unused parameter 'size' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3142:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3142 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3137:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3137 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3134:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3134 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3125:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3125 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3118:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3118 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3113:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3113 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3108:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3108 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3103:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3103 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3061:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3061 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3057:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3057 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3040:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3040 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2978:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2978 |             break;
      |             ^~~~~
29 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/gla.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  132 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:131:9: note: declared here
  131 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:67: warning: unused parameter 'size' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3142:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3142 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3137:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3137 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3134:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3134 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3125:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3125 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3118:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3118 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3113:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3113 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3108:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3108 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3103:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3103 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3061:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3061 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3057:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3057 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3040:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3040 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2978:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2978 |             break;
      |             ^~~~~
29 warnings generated when compiling for gfx90a.
6 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/gla.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  132 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:131:9: note: declared here
  131 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:67: warning: unused parameter 'size' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3142:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3142 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3137:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3137 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3134:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3134 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3125:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3125 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3118:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3118 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3113:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3113 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3108:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3108 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3103:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3103 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3061:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3061 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3057:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3057 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3040:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3040 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2978:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2978 |             break;
      |             ^~~~~
29 warnings generated when compiling for gfx90a.
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  132 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:131:9: note: declared here
  131 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:67: warning: unused parameter 'size' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3142:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3142 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3137:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3137 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3134:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3134 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3125:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3125 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3118:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3118 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3113:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3113 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3108:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3108 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3103:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3103 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3061:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3061 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3057:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3057 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3040:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3040 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2978:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2978 |             break;
      |             ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/gla.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
29 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:5:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:22:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
6 warnings generated when compiling for gfx1150.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  132 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:132:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:131:9: note: declared here
  131 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:52: warning: unused parameter 'buffer' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2837:67: warning: unused parameter 'size' [-Wunused-parameter]
 2837 | bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size) {
      |                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3142:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3142 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3137:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 3137 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3134:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3134 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3125:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3125 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3118:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3118 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3113:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3113 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3108:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3108 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3103:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3103 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3061:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3061 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3057:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3057 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:3040:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 3040 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/ggml-cuda.cu:2978:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 2978 |             break;
      |             ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/gla.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
29 warnings generated when compiling for host.
[ 32%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/im2col.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/im2col.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/im2col.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/im2col.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1151.
6 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/gla.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1012.
6 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/gla.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1030.
6 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/gla.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1201.
6 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/gla.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx900.
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/gla.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1102.
6 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/gla.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1150.
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/gla.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
6 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/gla.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1200.
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/gla.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx942.
6 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/gla.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 33%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmq.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmq.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmq.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmq.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu
6 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1010.
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1012.
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1030.
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1031.
6 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/im2col.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1035.
6 warnings generated when compiling for host.
[ 34%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmv.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmv.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmv.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmv.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmv.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1100.
6 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1102.
6 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1103.
6 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
6 warnings generated when compiling for gfx1031.
15 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1152.
6 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx900.
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx90a.
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx90a.
6 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:1:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cu:90:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   90 |             break;
      |             ^~~~~
6 warnings generated when compiling for gfx1151.
15 warnings generated when compiling for host.
[ 35%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmvq.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmvq.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmvq.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmvq.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:418:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  418 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
6 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmv.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 36%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/norm.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/norm.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/norm.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/norm.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1100.
293 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1101.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:418:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  418 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/norm.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 36%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/opt-step-adamw.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/opt-step-adamw.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/opt-step-adamw.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/opt-step-adamw.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
293 warnings generated when compiling for gfx1012.
6 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:418:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  418 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
6 warnings generated when compiling for gfx1200.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
161 warnings generated when compiling for gfx1030.
6 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/opt-step-adamw.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:418:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  418 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  1856 |  warning s  generated   when compiling for host .
       mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ 37%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/out-prod.cu.o
[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/out-prod.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/out-prod.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/out-prod.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cu
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
161 warnings generated when compiling for gfx1031.
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:418:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  418 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, s6 warnings generated when compiling for gfx1102.
tream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx900.
161 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:418:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  418 |             break;
      |             ^~~~~
6 warnings generated when compiling for gfx906.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/out-prod.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 38%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/pad.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/pad.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/pad.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/pad.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx1012.
161 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx1030.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:418:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  418 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
161 warnings generated when compiling for gfx1101.
8 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:418:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  418 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
8 warnings generated when compiling for gfx1151.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
161 warnings generated when compiling for gfx1102.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:418:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  418 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 82 warning>s< generated<< when compiling for bgfx90al.
ock_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:56: warning: comparison of integers of different signs: 'unsigned int' and 'int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                                             ~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pad.cu:17:35: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
   17 |     if (nidx < ne00 && blockIdx.y < ne01 && blockIdx.z < ne02*ne03) {
      |                        ~~~~~~~~~~ ^ ~~~~
8 warnings generated when compiling for host.
[ 39%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/pool2d.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/pool2d.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/pool2d.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/pool2d.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
161 warnings generated when compiling for gfx1103.
6 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:418:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  418 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
6 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1152.
161 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:418:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  418 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx942.
161 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/pool2d.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 40%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/quantize.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/quantize.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/quantize.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/quantize.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:418:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  418 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/quantize.cu:167:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  167 |             break;
      |             ^~~~~
15 warnings generated when compiling for host.
[ 41%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/rope.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/rope.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/rope.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/rope.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
293 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:418:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  418 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
6 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
293 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
6 warnings generated when compiling for gfx1201.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:418:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  418 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/rope.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 41%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/scale.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/scale.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/scale.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/scale.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1151.
293 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1152.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:418:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  418 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/scale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 42%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/softmax.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/softmax.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/softmax.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/softmax.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/softmax.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
293 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1150.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:418:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  418 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/softmax.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 43%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/sum.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/sum.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/sum.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/sum.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sum.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sum.cu:10:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sum.cu:10:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sum.cu:10:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sum.cu:10:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1031.
293 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sum.cu:10:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:418:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  418 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sum.cu:10:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(v/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.hx:,254 :v9y:,  warning: danonymous types declared in an anonymous union are an extension [-Wnested-anon-types]s
t, n  c254o | l s _ x ,   n r oswtsr_uxc,t  n{r
o      w| s        ^_
y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadId/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
x.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
6 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sum.cu:10:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sum.cu:10:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sum.cu:10:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sum.cu:10:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sum.cu:10:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sum.cu:10:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sum.cu:10:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sum.cu:10:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sum.cu:10:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sum.cu:10:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sum.cu:10:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sum.cu:10:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sum.cu:10:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sum.cu:10:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sum.cu:10:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 44%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/sumrows.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/sumrows.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/sumrows.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/sumrows.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1031.
293 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:418:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  418 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/sumrows.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 45%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/tsembd.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/tsembd.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/tsembd.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/tsembd.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
293 warnings generated when compiling for gfx90a.
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:418:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  418 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
6 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 45%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/unary.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/unary.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/unary.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/unary.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1012.
293 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:418:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  418 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 46%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/upscale.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/upscale.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/upscale.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/upscale.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
7 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
7 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
7 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
7 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
7 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
7 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
7 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
293 warnings generated when compiling for gfx942.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
7 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:57:34: warning: unused parameter 'nrows_x' [-Wunused-parameter]
   57 |     const int ncols_x, const int nrows_x, const int nrows_y, const int nrows_dst) {
      |                                  ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:418:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  418 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:209:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>' requested here
  209 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:216:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>' requested here
  216 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:223:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>' requested here
  223 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_1, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:230:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>' requested here
  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q8_0, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:237:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>' requested here
  237 |     mul_mat_vec_q_cuda<GGML_TYPE_Q8_0>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q2_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:244:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>' requested here
  244 |     mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:251:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>' requested here
  251 |     mul_mat_vec_q_cuda<GGML_TYPE_Q3_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q4_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:258:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>' requested here
  258 |     mul_mat_vec_q_cuda<GGML_TYPE_Q4_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q5_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:265:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>' requested here
  265 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q6_K, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:272:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>' requested here
  272 |     mul_mat_vec_q_cuda<GGML_TYPE_Q6_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:279:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>' requested here
  279 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:286:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>' requested here
  286 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ2_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:293:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>' requested here
  293 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ2_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_XXS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:300:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>' requested here
  300 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_XXS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:307:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>' requested here
  307 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 6>' requested here
  /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ1_M, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:314:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>' requested here
  314 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ1_M>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:321:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>' requested here
  321 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_NL>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_XS, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:328:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>' requested here
  328 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ4_XS>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 1>' requested here
  176 |             mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:179:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 2>' requested here
  179 |             mul_mat_vec_q<type, 2><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:185:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 4>' requested here
  185 |             mul_mat_vec_q<type, 4><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:188:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 5>' requested here
  188 |             mul_mat_vec_q<type, 5><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:191:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 6>' requested here
  191 |             mul_mat_vec_q<type, 6><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:194:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 7>' requested here
  194 |             mul_mat_vec_q<type, 7><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:197:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 8>' requested here
  197 |             mul_mat_vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, ncols_x, nrows_x, nrows_y, nrows_dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:335:5: note: in instantiation of function template specialization 'mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>' requested here
  335 |     mul_mat_vec_q_cuda<GGML_TYPE_IQ3_S>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
7 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
293 warnings generated when compiling for host.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
[ 47%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/wkv6.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/wkv6.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/wkv6.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/wkv6.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/wkv6.cu
7 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
7 warnings generated when compiling for gfx1151.
6 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
7 warnings generated when compiling for gfx1152.
6 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
7 warnings generated when compiling for gfx1200.
6 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
7 warnings generated when compiling for gfx1201.
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
7 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
7 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1100.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
7 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1101.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
7 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
6 warnings generated when compiling for gfx1102.
7 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
7 warnings generated when compiling for gfx942.
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu:22:37: warning: cast from 'const float *' to 'char *' drops const qualifier [-Wcast-qual]
   22 |     dst[index] = *(float *)((char *)x + i03 * nb03 + i02 * nb02 + i01 * nb01 + i00 * nb00);
      |                                     ^
7 warnings generated when compiling for host.
[ 48%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu
6 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
6 warnings generated when compiling for gfx1151.
64 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
6 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
64 warnings generated when compiling for gfx1030.
6 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
6 warnings generated when compiling for gfx900.
64 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
6/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
 warning/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
s generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
64 warnings generated when compiling for gfx1100.
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
6 warnings generated when compiling for gfx90a.
64 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
6 warnings generated when compiling for gfx942.
64 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/wkv6.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
6 warnings generated when compiling for host.
[ 49%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
64 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
61 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
64 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
61 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
64 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
61 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
64 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
61 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
64 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
61 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
64 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
61 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
64 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
61 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx906.
61 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx908.
61 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |  In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_       *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_ma*) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
x_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mIn file included from a/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cus:k3;:

In file included from       /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh| :                       ^2
:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24:/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh :warning: 704cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]:
5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  554 |      704  |         *f(l(ausihn_ta3tt2n__tc o*m)b i&nKeQ__rmeasxu_lstcsa<lDe,)  p&a=r afltlze_lm_abslko;c
k      s| >                       ^

      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh704::4915::9 :note:  in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested herenote: 
in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  704 |     491  |   f l a s h _ a tltanu_nccohm_bfiantet_nr<eDs,u lptasr<aDl,l epla_rballolcekls_>b(lcotcxk,s >d
s      t| ,    ^ 
fat/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuht:n505_:k5e:r nnote: ein instantiation of function template specialization 'launch_fattn<80, 1>' requested herel
, nwar  p505s | ,   c o llsa_upnecrh__bflaotctkn,< Dt,r upea,r atlrlueel)_;b
l      o| c        ^k
s>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, tr:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: ue, true);
      |         ^
cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scaleIn file included from )/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_f a&t=tn <ftDz, _pmaasrka;l
l      e| l                       ^_
blocks>(c/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuht:x704,: 5d:s tnote: ,in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here 
fattn  _704k | e r n e lf,l answha_raptst,n _ccoolmsb_ipneer__rbelsouclkt,s <tDr, upea,r atlrlueel)_;b
l      o| c        ^k
s>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
61 warnings generated when compiling for gfx1150.
64 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
61 warnings generated when compiling for gfx1151.
64 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
61 warnings generated when compiling for gfx1152.
64 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
61 warnings generated when compiling for gfx1200.
64 warnings generated when compiling for host.
[ 50%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
61 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h::298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
61 warningsIn file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
 generated when compiling for gfx900.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
61 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
61 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
61 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
61 warnings generated when compiling for gfx90a.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
61 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
61 warnings generated when compiling for host.
[ 50%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
64 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
64 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
64 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1035.
64 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1100.
64 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1101.
64 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1102.
64 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1103.
64 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 | In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
        struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
64 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
64 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
64 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for host.
[ 51%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
64 warnings generated when compiling for gfx1201.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
58 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx900.
58 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx906.
58 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
64 warnings generated when compiling for gfx908.
58 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh  554 |         *((ui:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
nt32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |    In file included from  /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
    *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^

  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &=*) &KQ_m ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flaaxsh__scatatlne_)c o&m=bi nftez__rmeassukl;t
s      <| D                       ^,
 parallel/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh_:b704l:o5c:k snote: >in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here

      |     ^
  /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh704: | 491 : 9 :  fnote: lin instantiation of function template specialization 'launch_fattn<96, 2>' requested herea
sh_attn  _491c | o m b i n e _ r elsauulntcsh<_Df,a tptanr<aDl,l epla_rballolcekls_>b
l      o| c    ^k
s>(c/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuht:x505,: 5d: snote: tin instantiation of function template specialization 'launch_fattn<256, 1>' requested here, 
fattn_k  e505r | n e l ,  lnawuanrcphs_,f actotlns<_Dp,e rp_abrlaolclke,l _tbrluoec,k st>r(ucet)x;,
       d| s        ^t
, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
58 warnings generated when compiling for gfx1035.
64 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
58 warnings generated when compiling for gfx1100.
64 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
58 warnings generated when compiling for gfx1101.
64 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<80, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<80, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<80, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<80, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<112, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<112, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<112, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<112, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
58 warnings generated when compiling for gfx1102.
64 warnings generated when compiling for host.
[ 52%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq1_s.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq1_s.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq1_s.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq1_s.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
58 warnings generated when compiling for gfx1103.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
58 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
58 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
58 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
58 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
58 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
58 warnings generated when compiling for gfx900.
78 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
58 warnings generated when compiling for gfx906.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
58 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
58 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
58 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
58 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:14:35: warning: unused parameter 'Q' [-Wunused-parameter]
   14 |         const char * __restrict__ Q,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35: warning: unused parameter 'K' [-Wunused-parameter]
   15 |         const char * __restrict__ K,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:16:35: warning: unused parameter 'V' [-Wunused-parameter]
   16 |         const char * __restrict__ V,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:17:35: warning: unused parameter 'mask' [-Wunused-parameter]
   17 |         const char * __restrict__ mask,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:18:35: warning: unused parameter 'dst' [-Wunused-parameter]
   18 |         float      * __restrict__ dst,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:19:35: warning: unused parameter 'dst_meta' [-Wunused-parameter]
   19 |         float2     * __restrict__ dst_meta,
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:20:21: warning: unused parameter 'scale' [-Wunused-parameter]
   20 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:21:21: warning: unused parameter 'max_bias' [-Wunused-parameter]
   21 |         const float max_bias,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:22:21: warning: unused parameter 'm0' [-Wunused-parameter]
   22 |         const float m0,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:23:21: warning: unused parameter 'm1' [-Wunused-parameter]
   23 |         const float m1,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:24:24: warning: unused parameter 'n_head_log2' [-Wunused-parameter]
   24 |         const uint32_t n_head_log2,
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:25:21: warning: unused parameter 'logit_softcap' [-Wunused-parameter]
   25 |         const float logit_softcap,
      |                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   26 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:27:19: warning: unused parameter 'ne01' [-Wunused-parameter]
   27 |         const int ne01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:28:19: warning: unused parameter 'ne02' [-Wunused-parameter]
   28 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:29:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   29 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:30:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   30 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:31:19: warning: unused parameter 'ne11' [-Wunused-parameter]
   31 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:32:19: warning: unused parameter 'ne12' [-Wunused-parameter]
   32 |         const int ne12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:33:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   33 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:34:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   34 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:35:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   35 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:36:19: warning: unused parameter 'nb01' [-Wunused-parameter]
   36 |         const int nb01,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:37:19: warning: unused parameter 'nb02' [-Wunused-parameter]
   37 |         const int nb02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:38:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   38 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:39:19: warning: unused parameter 'nb11' [-Wunused-parameter]
   39 |         const int nb11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:40:19: warning: unused parameter 'nb12' [-Wunused-parameter]
   40 |         const int nb12,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:41:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   41 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:42:19: warning: unused parameter 'nb21' [-Wunused-parameter]
   42 |         const int nb21,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:43:19: warning: unused parameter 'nb22' [-Wunused-parameter]
   43 |         const int nb22,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   44 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:45:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   45 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   46 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:47:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   47 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:48:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   48 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<64, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<96, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<128, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:476:9: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:491:9: note: in instantiation of function template specialization 'launch_fattn<256, 2>' requested here
  491 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:505:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  505 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
58 warnings generated when compiling for host.
[ 53%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_s.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_s.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_s.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_s.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1012.
78 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1100.
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A,/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
 const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh :i2691n:t36 :&  warning: ncomparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]e
00, cons t2691  | i n t   &   n e 0i1f,  (ciotn s!t=  ibnlto c&k Isdtxr.ixd e|0|1 ,j tc o!n=s tb lionctk I&d xn.ey1)0 ,{ 
c      o| n                                ~~ ^  ~~~~~~~~~~s
t int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
78 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
78 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 54%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 54%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
78 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 55%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq3_s.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq3_s.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq3_s.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq3_s.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx942.
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 56%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
78 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_S, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 57%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
78 warnings generated when compiling for gfx1010.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ3_XXS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 58%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
78 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1152.
78 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_NL, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 59%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q2_k.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q2_k.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q2_k.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q2_k.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
 1022 |     for (int k01 = 0; k01 < WARP_SIZE; k01 += QR2_K*VDR_Q2_K_Q8_1_MMQ) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
78 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ4_XS, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 59%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q3_k.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q3_k.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q3_k.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q3_k.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
98 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
 1022 |     for (int k01 = 0; k01 < WARP_SIZE; k01 += QR2_K*VDR_Q2_K_Q8_1_MMQ) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
78 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
118 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 60%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_0.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_0.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_0.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_0.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1012.
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
78 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
 1022 |     for (int k01 = 0; k01 < WARP_SIZE; k01 += QR2_K*VDR_Q2_K_Q8_1_MMQ) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
134 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 61%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_1.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_1.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_1.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_1.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
 1022 |     for (int k01 = 0; k01 < WARP_SIZE; k01 += QR2_K*VDR_Q2_K_Q8_1_MMQ) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
78 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
134 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
 1022 |     for (int k01 = 0; k01 < WARP_SIZE; k01 += QR2_K*VDR_Q2_K_Q8_1_MMQ) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
78 warnings generated when compiling for gfx1012.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
134 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
 1022 |     for (int k01 = 0; k01 < WARP_SIZE; k01 += QR2_K*VDR_Q2_K_Q8_1_MMQ) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
78 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
110 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 62%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_k.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_k.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_k.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_k.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
 1022 |     for (int k01 = 0; k01 < WARP_SIZE; k01 += QR2_K*VDR_Q2_K_Q8_1_MMQ) {
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
134 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 63%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_0.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_0.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_0.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_0.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
78 warnings generated when compiling for gfx1151.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
78 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
78 warnings generated when compiling for gfx1200.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 63%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_1.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_1.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_1.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_1.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 64%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_k.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_k.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_k.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_k.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 65%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q6_k.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q6_k.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q6_k.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q6_k.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 66%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q8_0.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q8_0.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q8_0.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q8_0.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1201.
78 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, st/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.hre:a213m:)9;:
       warning: | anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]            ^

  213/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh: | 2691 : 16 :   warning:  comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare] 
 stru ct2691  | { 
        |          ^ 
   if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q6_K, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for host.
[ 67%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
60 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
78 warnings generated when compiling for gfx942.
60 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:866:99: warning: unused parameter 'k00' [-Wunused-parameter]
  866 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2852:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 2852 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 16, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2855:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 16>' requested here
 2855 |             launch_mul_mat_q<type,  16>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 24, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2858:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 24>' requested here
 2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 32, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2861:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 32>' requested here
 2861 |             launch_mul_mat_q<type,  32>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softc/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuha:p2691>:;36
:       | warning:                                   ^comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]

/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13:  2691note:  | in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here 
         i330f |   ( i t   ! =   b l o c kgIgdmxl._xc u|d|a _jftl a!s=h _baltotcnk_Iedxxt._yv)e c{_
f      1| 6                                ~~ ^  ~~~~~~~~~~_
case_im/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuhp:l2805<:D9,:  cnote: oin instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, false>' requested herel
s_per_bl o2805c | k ,   p a r a l lmeull__bmlaotc_kqs_,s ttryepaem__Kk,_ ftiyxpuep_<Vt,y pues,e _mlmoqg_ixt,_ sMoMfQt_cNaWpA>R(PcSt,x ,n edesdt_)c;h
e      c| k            ^>
<<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 40, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2864:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 40>' requested here
 2864 |             launch_mul_mat_q<type,  40>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 48, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2867:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 48>' requested here
 2867 |             launch_mul_mat_q<type,  48>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 56, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2870:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 56>' requested here
 2870 |             launch_mul_mat_q<type,  56>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2873:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
 2873 |             launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 72, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2876:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 72>' requested here
 2876 |             launch_mul_mat_q<type,  72>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 80, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2879:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 80>' requested here
 2879 |             launch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 88, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2882:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 88>' requested here
 2882 |             launch_mul_mat_q<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 96, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 96>' requested here
 2885 |             launch_mul_mat_q<type,  96>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 104, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2888:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 104>' requested here
 2888 |             launch_mul_mat_q<type, 104>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 112, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2891:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 112>' requested here
 2891 |             launch_mul_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2813:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 120, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2894:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 120>' requested here
 2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2897:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here
 2897 |             launch_mul_mat_q<type, 128>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh_S:I2691Z:E36):  {warning: 
comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]      | 
                             ~~ ^ ~~~~~~~~~~~~~
 2691 |    /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh : 303 : 35 :i fnote:  (in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested herei
t != blo  c303k | Id x . x  f|a|t tjnt_ k!e=rn ebll_otc kfIadtxt.ny_)k e{r
n      e| l                                ~~ ^  ~~~~~~~~~~ 
= flash/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:_2813a:t9t:n _note: vin instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 128, 8, true>' requested heree
c_ext_f16 2813<D | ,   c o l s _ p emru_lb_lmoactk_,q _psatrraelalme_lk__bflioxcukps<,t ytpyep,e _mKm,q _txy,p eM_MVQ,_ NuWsAeR_PlSo,g inte_esdo_fcthceacpk>>;<
<      | <b                                  ^l
ock_/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuhn:u346m:s13_:x ynote: _in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested heret
iling, bl  o346c | k _ d i m s ,   0 ,   s tgrgemalm_>cu>d>a
_      f|         ^la
sh_/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuha:t2897t:n13_:e xnote: tin instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 128>' requested here_
vec_f 128976 | _ c a s e _ i m p l < D ,l acuonlcsh__pmeurl__bmlaotck_,q <ptayraplel,el _1b2l8o>c(ksc,t txy,p e_aKr, gtsy,p es_tVr,e ausme)_;l
o      g| it            ^_
softcap>/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh(:c2691tx:,16 :d stwarning: )comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare];

      |             ^
 2691 | /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh : 142 : 33 :   warning:  comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]i
f (it   !=142  | b l o c k Id x . x   | |  fjot r! =( ibnlto cik0Id x=.y )0 ;{ 
i      0|              ~~ ^  ~~~~~~~~~~<
 D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
78 warnings generated when compiling for host.
[ 68%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1151.
60 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1102.
60 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_cas:281:e9_:i mwarning: planonymous types declared in an anonymous union are an extension [-Wnested-anon-types]<
D, co  l281s | _ p er _ b l o c ks,t rpuacrta l{l
e      l| _        ^b
locks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for host.
[ 68%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1201.
60 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1031.
60 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q4_0, GGML_TYPE_Q4_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for host.
[ 69%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1031.
60 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
60 warnings generated when compiling for gfx942.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:333:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  333 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:343:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  343 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:346:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  346 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:356:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  356 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:359:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  359 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:369:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  369 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:372:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  372 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:384:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  384 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for host.
[ 70%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
60 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
34 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
60 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
34 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
60 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
34 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
60 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
34 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
60 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
34 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
60 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
34 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
60 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
34 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
60 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
34 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
60 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
34 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
60 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
34 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1152.
60 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
34 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
60 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
34 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
60 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:311:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  311 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:321:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  321 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:324:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 2, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  324 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:334:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  334 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:337:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  337 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:347:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  347 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:350:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  350 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:362:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here
  362 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:142:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  142 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
60 warnings generated when compiling for host.
[ 71%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
34 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx906.
34 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 128>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 128>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 128>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 128>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 128>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for host.
[ 72%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu
34 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1100.
34 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1151.
34 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03      ,| 
                             ^| 
                  ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19:/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh :warning: 704unused parameter 'ne10' [-Wunused-parameter]:
5:    note: 25in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here | 
         704  | c o n s tf liansth _naet1t0n,_
c      o| m                  ^b
in/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuhe:_28r:e19s:u lwarning: tunused parameter 'ne13' [-Wunused-parameter]s
<D,    28p | a r a l l e l _ bclooncskts >i
n      t|      ^n
e13/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh,:
306      :| 5                  ^:
 note: /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuhin instantiation of function template specialization 'launch_fattn<64, 4>' requested here:
29:19: warning: unused parameter 'ne31' [-Wunused-parameter]  
306 |      29  |   l a u n c h _ fcaotntsnt< Di,n tp anrea3l1l,e
l      _| b                  ^l
oc/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuhk:s30>:(19c:t xwarning: ,unused parameter 'nb31' [-Wunused-parameter] 
dst   ,30  | f a t t n _ k e rcnoenls,t  niwnatr pnsb,3 1c,o
l      s| _                  ^p
er_/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuhb:l33o:c19k:,  warning: nunused parameter 'nb03' [-Wunused-parameter]e
ed_   f331 | 6 _ K ,   n e e dc_ofn1s6t_ Vi)n;t
       n| b    ^0
3,
      /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh| :                  ^330
:13/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:: 36note: :in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here19
: warning: unused parameter 'nb13' [-Wunused-parameter]
   36   | 330  |               c o n s t  gignmtl _ncbu1d3a,_
f      l| a                  ^s
h_a/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuht:t39n:_19e:x twarning: _unused parameter 'nb23' [-Wunused-parameter]v
ec_   f391 | 6 _ c a s e _ i mcpoln<sDt,  icnotl nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
s_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1152.
34 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1200.
34 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, t/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuhy:p554e:_24K:,  warning: tcast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]y
pe_V,   u554s | e _ l o g i t _ s*o(f(tucianpt>3;2
_      t|                                    ^*
) &/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuhK:Q330_:m13a:x _note: sin instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested herec
ale) &=   f330t | z _ m a s k ; 
        |                         ^ 
 ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
34 warnings generated when compiling for gfx1201.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx90a.
34 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 64>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 64>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 64>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 64>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 64>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for host.
[ 72%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:483:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<__half, 256>' requested here
  483 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:482:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<__half, 256>' requested here
  482 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:481:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<__half, 256>' requested here
  481 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:480:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<__half, 256>' requested here
  480 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<half, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:479:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<__half, 256>' requested here
  479 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<half, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<__half2>' requested here
  130 |                 quantize_q8_1_to_shared<half2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:303:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f16<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  303 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f16<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:330:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  330 |             ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:306:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  306 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:381:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f16_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  381 |         ggml_cuda_flash_attn_ext_vec_f16_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for host.
[ 73%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
34 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
34 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 128>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 128>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 128>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 128>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 128>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<128, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for host.
[ 74%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_HIP_NO_VMM -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_HIP -DUSE_PROF_API=1 -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -D__HIP_PLATFORM_AMD__=1 -Dggml_hip_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -x hip --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu.o -MF CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu.o.d -o CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx900.
34 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx900.
34 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 256>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 256>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 256>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 256>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 256>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<256, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<256, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<256, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for host.
34 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:21:19: warning: unused parameter 'ne00' [-Wunused-parameter]
   21 |         const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:24:19: warning: unused parameter 'ne03' [-Wunused-parameter]
   24 |         const int ne03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:25:19: warning: unused parameter 'ne10' [-Wunused-parameter]
   25 |         const int ne10,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:28:19: warning: unused parameter 'ne13' [-Wunused-parameter]
   28 |         const int ne13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:29:19: warning: unused parameter 'ne31' [-Wunused-parameter]
   29 |         const int ne31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:30:19: warning: unused parameter 'nb31' [-Wunused-parameter]
   30 |         const int nb31,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:33:19: warning: unused parameter 'nb03' [-Wunused-parameter]
   33 |         const int nb03,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:36:19: warning: unused parameter 'nb13' [-Wunused-parameter]
   36 |         const int nb13,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:39:19: warning: unused parameter 'nb23' [-Wunused-parameter]
   39 |         const int nb23,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:40:19: warning: unused parameter 'ne0' [-Wunused-parameter]
   40 |         const int ne0,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:41:19: warning: unused parameter 'ne1' [-Wunused-parameter]
   41 |         const int ne1,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:42:19: warning: unused parameter 'ne2' [-Wunused-parameter]
   42 |         const int ne2,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:43:19: warning: unused parameter 'ne3' [-Wunused-parameter]
   43 |         const int ne3) {
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:247:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  247 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:494:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q8_0<float, 64>' requested here
  494 |         type_K == GGML_TYPE_Q8_0 ? vec_dot_fattn_vec_KQ_q8_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:196:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  196 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:493:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_1<float, 64>' requested here
  493 |         type_K == GGML_TYPE_Q5_1 ? vec_dot_fattn_vec_KQ_q5_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:149:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  149 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:492:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q5_0<float, 64>' requested here
  492 |         type_K == GGML_TYPE_Q5_0 ? vec_dot_fattn_vec_KQ_q5_0<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:105:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  105 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:491:36: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_1<float, 64>' requested here
  491 |         type_K == GGML_TYPE_Q4_1 ? vec_dot_fattn_vec_KQ_q4_1<float, D> :
      |                                    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:65:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
   65 |     for (int k_KQ_0 = 0; k_KQ_0 < D/sizeof(int); k_KQ_0 += WARP_SIZE) {
      |                          ~~~~~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:490:39: note: in instantiation of function template specialization 'vec_dot_fattn_vec_KQ_q4_0<float, 64>' requested here
  490 |     return type_K == GGML_TYPE_Q4_0 ? vec_dot_fattn_vec_KQ_q4_0<float, D> :
      |                                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:130:17: note: in instantiation of function template specialization 'quantize_q8_1_to_shared<HIP_vector_type<float, 2>>' requested here
  130 |                 quantize_q8_1_to_shared<float2>(Q_f + 4*i0, scale, tmp_q_i32, tmp_q_ds);
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:284:35: note: in instantiation of function template specialization 'flash_attn_vec_ext_f32<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  284 |     fattn_kernel_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 4>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:308:13: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 1, 4, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  308 |             ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<64, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:287:5: note: in instantiation of function template specialization 'launch_fattn<64, 1>' requested here
  287 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, need_f16_K, need_f16_V);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:359:9: note: in instantiation of function template specialization 'ggml_cuda_flash_attn_ext_vec_f32_case_impl<64, 8, 1, GGML_TYPE_F16, GGML_TYPE_F16, false>' requested here
  359 |         ggml_cuda_flash_attn_ext_vec_f32_case_impl<D, cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |         ^
34 warnings generated when compiling for host.
[ 75%] Linking CXX shared library ../../../bin/libggml-hip.so
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/cmake -E cmake_link_script CMakeFiles/ggml-hip.dir/link.txt --verbose=1
/usr/bin/hipcc -fPIC -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -Xlinker --dependency-file=CMakeFiles/ggml-hip.dir/link.d -Wl,-z,relro -Wl,--as-needed  -Wl,-z,pack-relative-relocs -Wl,-z,now -Wl,-z,now   -Wl,--build-id=sha1 -specs=/usr/lib/rpm/redhat/redhat-package-notes -shared -Wl,-soname,libggml-hip.so.b4580 -o ../../../bin/libggml-hip.so.b4580 "CMakeFiles/ggml-hip.dir/__/ggml-cuda/acc.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/arange.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/argmax.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/argsort.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/binbcast.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/clamp.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/concat.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/conv-transpose-1d.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/convert.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/count-equal.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/cpy.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/cross-entropy-loss.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/diagmask.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn-tile-f16.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn-tile-f32.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/getrows.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/ggml-cuda.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/gla.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/im2col.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmq.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmv.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmvq.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/norm.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/opt-step-adamw.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/out-prod.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/pad.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/pool2d.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/quantize.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/rope.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/scale.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/softmax.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/sum.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/sumrows.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/tsembd.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/unary.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/upscale.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/wkv6.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq1_s.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_s.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq3_s.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q2_k.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q3_k.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-iclang++: warning: argument unused during compilation: '-Xarch_host -fstack-protector-strong' [-Wunused-command-line-argument]
clang++: warning: argument unused during compilation: '-Xarch_host -fcf-protection' [-Wunused-command-line-argument]
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]
nstance-q4_0.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_1.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_k.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_0.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_1.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_k.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q6_k.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q8_0.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu.o" "CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu.o"  ../../../bin/libggml-base.so.b4580 /usr/lib64/libhipblas.so.2.3 --hip-link --offload-arch=gfx900 --offload-arch=gfx906:xnack- --offload-arch=gfx908:xnack- --offload-arch=gfx90a:xnack+ --offload-arch=gfx90a:xnack- --offload-arch=gfx942 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1200 --offload-arch=gfx1201 /usr/lib64/librocblas.so.4.3 /usr/lib64/libamdhip64.so.6.3.42134
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/ggml-hip && /usr/bin/cmake -E cmake_symlink_library ../../../bin/libggml-hip.so.b4580 ../../../bin/libggml-hip.so.b4580 ../../../bin/libggml-hip.so
gmake[2]: Leaving directory '/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build'
[ 75%] Built target ggml-hip
/usr/bin/gmake  -f ggml/src/CMakeFiles/ggml.dir/build.make ggml/src/CMakeFiles/ggml.dir/depend
gmake[2]: Entering directory '/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build'
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build && /usr/bin/cmake -E cmake_depends "Unix Makefiles" /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580 /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src/CMakeFiles/ggml.dir/DependInfo.cmake "--color="
gmake[2]: Leaving directory '/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build'
/usr/bin/gmake  -f ggml/src/CMakeFiles/ggml.dir/build.make ggml/src/CMakeFiles/ggml.dir/build
gmake[2]: Entering directory '/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build'
[ 75%] Building CXX object ggml/src/CMakeFiles/ggml.dir/ggml-backend-reg.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_BUILD -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -std=gnu++17 -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT ggml/src/CMakeFiles/ggml.dir/ggml-backend-reg.cpp.o -MF CMakeFiles/ggml.dir/ggml-backend-reg.cpp.o.d -o CMakeFiles/ggml.dir/ggml-backend-reg.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-backend-reg.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[ 76%] Linking CXX shared library ../../bin/libggml.so
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src && /usr/bin/cmake -E cmake_link_script CMakeFiles/ggml.dir/link.txt --verbose=1
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
clang++: warning: argument unused during compilation: '-Xarch_host -fstack-protector-strong' [-Wunused-command-line-argument]
clang++: warning: argument unused during compilation: '-Xarch_host -fcf-protection' [-Wunused-command-line-argument]
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]
/usr/bin/hipcc -fPIC -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -Xlinker --dependency-file=CMakeFiles/ggml.dir/link.d -Wl,-z,relro -Wl,--as-needed  -Wl,-z,pack-relative-relocs -Wl,-z,now -Wl,-z,now   -Wl,--build-id=sha1 -specs=/usr/lib/rpm/redhat/redhat-package-notes -shared -Wl,-soname,libggml.so.b4580 -o ../../bin/libggml.so.b4580 "CMakeFiles/ggml.dir/ggml-backend-reg.cpp.o"  -ldl ../../bin/libggml-cpu.so.b4580 ../../bin/libggml-hip.so.b4580 ../../bin/libggml-base.so.b4580
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/ggml/src && /usr/bin/cmake -E cmake_symlink_library ../../bin/libggml.so.b4580 ../../bin/libggml.so.b4580 ../../bin/libggml.so
gmake[2]: Leaving directory '/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build'
[ 76%] Built target ggml
/usr/bin/gmake  -f src/CMakeFiles/llama.dir/build.make src/CMakeFiles/llama.dir/depend
gmake[2]: Entering directory '/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build'
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build && /usr/bin/cmake -E cmake_depends "Unix Makefiles" /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580 /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/src /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/src/CMakeFiles/llama.dir/DependInfo.cmake "--color="
gmake[2]: Leaving directory '/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build'
/usr/bin/gmake  -f src/CMakeFiles/llama.dir/build.make src/CMakeFiles/llama.dir/build
gmake[2]: Entering directory '/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build'
[ 77%] Building CXX object src/CMakeFiles/llama.dir/llama.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/src && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dllama_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../include -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../common -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT src/CMakeFiles/llama.dir/llama.cpp.o -MF CMakeFiles/llama.dir/llama.cpp.o.d -o CMakeFiles/llama.dir/llama.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/llama.cpp
[ 78%] Building CXX object src/CMakeFiles/llama.dir/llama-adapter.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/src && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dllama_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../include -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../common -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT src/CMakeFiles/llama.dir/llama-adapter.cpp.o -MF CMakeFiles/llama.dir/llama-adapter.cpp.o.d -o CMakeFiles/llama.dir/llama-adapter.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/llama-adapter.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[ 78%] Building CXX object src/CMakeFiles/llama.dir/llama-arch.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/src && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dllama_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../include -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../common -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT src/CMakeFiles/llama.dir/llama-arch.cpp.o -MF CMakeFiles/llama.dir/llama-arch.cpp.o.d -o CMakeFiles/llama.dir/llama-arch.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/llama-arch.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[ 79%] Building CXX object src/CMakeFiles/llama.dir/llama-batch.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/src && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dllama_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../include -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../common -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT src/CMakeFiles/llama.dir/llama-batch.cpp.o -MF CMakeFiles/llama.dir/llama-batch.cpp.o.d -o CMakeFiles/llama.dir/llama-batch.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/llama-batch.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[ 80%] Building CXX object src/CMakeFiles/llama.dir/llama-chat.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/src && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dllama_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../include -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../common -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT src/CMakeFiles/llama.dir/llama-chat.cpp.o -MF CMakeFiles/llama.dir/llama-chat.cpp.o.d -o CMakeFiles/llama.dir/llama-chat.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/llama-chat.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[ 81%] Building CXX object src/CMakeFiles/llama.dir/llama-context.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/src && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dllama_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../include -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../common -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT src/CMakeFiles/llama.dir/llama-context.cpp.o -MF CMakeFiles/llama.dir/llama-context.cpp.o.d -o CMakeFiles/llama.dir/llama-context.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/llama-context.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[ 82%] Building CXX object src/CMakeFiles/llama.dir/llama-grammar.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/src && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dllama_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../include -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../common -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT src/CMakeFiles/llama.dir/llama-grammar.cpp.o -MF CMakeFiles/llama.dir/llama-grammar.cpp.o.d -o CMakeFiles/llama.dir/llama-grammar.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/llama-grammar.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[ 82%] Building CXX object src/CMakeFiles/llama.dir/llama-hparams.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/src && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dllama_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../include -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../common -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT src/CMakeFiles/llama.dir/llama-hparams.cpp.o -MF CMakeFiles/llama.dir/llama-hparams.cpp.o.d -o CMakeFiles/llama.dir/llama-hparams.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/llama-hparams.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[ 83%] Building CXX object src/CMakeFiles/llama.dir/llama-impl.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/src && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dllama_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../include -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../common -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT src/CMakeFiles/llama.dir/llama-impl.cpp.o -MF CMakeFiles/llama.dir/llama-impl.cpp.o.d -o CMakeFiles/llama.dir/llama-impl.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/llama-impl.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[ 84%] Building CXX object src/CMakeFiles/llama.dir/llama-kv-cache.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/src && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dllama_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../include -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../common -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT src/CMakeFiles/llama.dir/llama-kv-cache.cpp.o -MF CMakeFiles/llama.dir/llama-kv-cache.cpp.o.d -o CMakeFiles/llama.dir/llama-kv-cache.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/llama-kv-cache.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[ 85%] Building CXX object src/CMakeFiles/llama.dir/llama-mmap.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/src && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dllama_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../include -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../common -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT src/CMakeFiles/llama.dir/llama-mmap.cpp.o -MF CMakeFiles/llama.dir/llama-mmap.cpp.o.d -o CMakeFiles/llama.dir/llama-mmap.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/llama-mmap.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[ 86%] Building CXX object src/CMakeFiles/llama.dir/llama-model-loader.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/src && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dllama_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../include -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../common -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT src/CMakeFiles/llama.dir/llama-model-loader.cpp.o -MF CMakeFiles/llama.dir/llama-model-loader.cpp.o.d -o CMakeFiles/llama.dir/llama-model-loader.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/llama-model-loader.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[ 87%] Building CXX object src/CMakeFiles/llama.dir/llama-model.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/src && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dllama_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../include -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../common -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT src/CMakeFiles/llama.dir/llama-model.cpp.o -MF CMakeFiles/llama.dir/llama-model.cpp.o.d -o CMakeFiles/llama.dir/llama-model.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/llama-model.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[ 87%] Building CXX object src/CMakeFiles/llama.dir/llama-quant.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/src && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dllama_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../include -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../common -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT src/CMakeFiles/llama.dir/llama-quant.cpp.o -MF CMakeFiles/llama.dir/llama-quant.cpp.o.d -o CMakeFiles/llama.dir/llama-quant.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/llama-quant.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[ 88%] Building CXX object src/CMakeFiles/llama.dir/llama-sampling.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/src && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dllama_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../include -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../common -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT src/CMakeFiles/llama.dir/llama-sampling.cpp.o -MF CMakeFiles/llama.dir/llama-sampling.cpp.o.d -o CMakeFiles/llama.dir/llama-sampling.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/llama-sampling.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[ 89%] Building CXX object src/CMakeFiles/llama.dir/llama-vocab.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/src && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dllama_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../include -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../common -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT src/CMakeFiles/llama.dir/llama-vocab.cpp.o -MF CMakeFiles/llama.dir/llama-vocab.cpp.o.d -o CMakeFiles/llama.dir/llama-vocab.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/llama-vocab.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[ 90%] Building CXX object src/CMakeFiles/llama.dir/unicode.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/src && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dllama_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../include -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../common -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT src/CMakeFiles/llama.dir/unicode.cpp.o -MF CMakeFiles/llama.dir/unicode.cpp.o.d -o CMakeFiles/llama.dir/unicode.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/unicode.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[ 91%] Building CXX object src/CMakeFiles/llama.dir/unicode-data.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/src && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dllama_EXPORTS -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../include -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../common -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT src/CMakeFiles/llama.dir/unicode-data.cpp.o -MF CMakeFiles/llama.dir/unicode-data.cpp.o.d -o CMakeFiles/llama.dir/unicode-data.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/unicode-data.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[ 92%] Linking CXX shared library ../bin/libllama.so
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/src && /usr/bin/cmake -E cmake_link_script CMakeFiles/llama.dir/link.txt --verbose=1
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
clang++: warning: argument unused during compilation: '-Xarch_host -fstack-protector-strong' [-Wunused-command-line-argument]
clang++: warning: argument unused during compilation: '-Xarch_host -fcf-protection' [-Wunused-command-line-argument]
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]
/usr/bin/hipcc -fPIC -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -Xlinker --dependency-file=CMakeFiles/llama.dir/link.d -Wl,-z,relro -Wl,--as-needed  -Wl,-z,pack-relative-relocs -Wl,-z,now -Wl,-z,now   -Wl,--build-id=sha1 -specs=/usr/lib/rpm/redhat/redhat-package-notes -shared -Wl,-soname,libllama.so.b4580 -o ../bin/libllama.so.b4580 CMakeFiles/llama.dir/llama.cpp.o "CMakeFiles/llama.dir/llama-adapter.cpp.o" "CMakeFiles/llama.dir/llama-arch.cpp.o" "CMakeFiles/llama.dir/llama-batch.cpp.o" "CMakeFiles/llama.dir/llama-chat.cpp.o" "CMakeFiles/llama.dir/llama-context.cpp.o" "CMakeFiles/llama.dir/llama-grammar.cpp.o" "CMakeFiles/llama.dir/llama-hparams.cpp.o" "CMakeFiles/llama.dir/llama-impl.cpp.o" "CMakeFiles/llama.dir/llama-kv-cache.cpp.o" "CMakeFiles/llama.dir/llama-mmap.cpp.o" "CMakeFiles/llama.dir/llama-model-loader.cpp.o" "CMakeFiles/llama.dir/llama-model.cpp.o" "CMakeFiles/llama.dir/llama-quant.cpp.o" "CMakeFiles/llama.dir/llama-sampling.cpp.o" "CMakeFiles/llama.dir/llama-vocab.cpp.o" CMakeFiles/llama.dir/unicode.cpp.o "CMakeFiles/llama.dir/unicode-data.cpp.o"  ../bin/libggml.so.b4580 ../bin/libggml-cpu.so.b4580 ../bin/libggml-hip.so.b4580 ../bin/libggml-base.so.b4580
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/src && /usr/bin/cmake -E cmake_symlink_library ../bin/libllama.so.b4580 ../bin/libllama.so.b4580 ../bin/libllama.so
gmake[2]: Leaving directory '/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build'
[ 92%] Built target llama
/usr/bin/gmake  -f common/CMakeFiles/common.dir/build.make common/CMakeFiles/common.dir/depend
gmake[2]: Entering directory '/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build'
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build && /usr/bin/cmake -E cmake_depends "Unix Makefiles" /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580 /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/common /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/common /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/common/CMakeFiles/common.dir/DependInfo.cmake "--color="
gmake[2]: Leaving directory '/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build'
/usr/bin/gmake  -f common/CMakeFiles/common.dir/build.make common/CMakeFiles/common.dir/build
gmake[2]: Entering directory '/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build'
[ 93%] Building CXX object common/CMakeFiles/common.dir/arg.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/common && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/common/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../include -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../common -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT common/CMakeFiles/common.dir/arg.cpp.o -MF CMakeFiles/common.dir/arg.cpp.o.d -o CMakeFiles/common.dir/arg.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/common/arg.cpp
[ 94%] Building CXX object common/CMakeFiles/common.dir/common.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/common && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/common/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../include -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../common -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT common/CMakeFiles/common.dir/common.cpp.o -MF CMakeFiles/common.dir/common.cpp.o.d -o CMakeFiles/common.dir/common.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/common/common.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[ 95%] Building CXX object common/CMakeFiles/common.dir/console.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/common && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/common/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../include -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../common -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT common/CMakeFiles/common.dir/console.cpp.o -MF CMakeFiles/common.dir/console.cpp.o.d -o CMakeFiles/common.dir/console.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/common/console.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[ 95%] Building CXX object common/CMakeFiles/common.dir/json-schema-to-grammar.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/common && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/common/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../include -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../common -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT common/CMakeFiles/common.dir/json-schema-to-grammar.cpp.o -MF CMakeFiles/common.dir/json-schema-to-grammar.cpp.o.d -o CMakeFiles/common.dir/json-schema-to-grammar.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/common/json-schema-to-grammar.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[ 96%] Building CXX object common/CMakeFiles/common.dir/log.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/common && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/common/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../include -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../common -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT common/CMakeFiles/common.dir/log.cpp.o -MF CMakeFiles/common.dir/log.cpp.o.d -o CMakeFiles/common.dir/log.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/common/log.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[ 97%] Building CXX object common/CMakeFiles/common.dir/ngram-cache.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/common && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/common/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../include -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../common -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT common/CMakeFiles/common.dir/ngram-cache.cpp.o -MF CMakeFiles/common.dir/ngram-cache.cpp.o.d -o CMakeFiles/common.dir/ngram-cache.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/common/ngram-cache.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[ 98%] Building CXX object common/CMakeFiles/common.dir/sampling.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/common && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/common/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../include -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../common -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT common/CMakeFiles/common.dir/sampling.cpp.o -MF CMakeFiles/common.dir/sampling.cpp.o.d -o CMakeFiles/common.dir/sampling.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/common/sampling.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[ 99%] Building CXX object common/CMakeFiles/common.dir/speculative.cpp.o
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/common && /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/common/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/. -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../include -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/src/../common -I/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/../include -O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -DNDEBUG -fPIC -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT common/CMakeFiles/common.dir/speculative.cpp.o -MF CMakeFiles/common.dir/speculative.cpp.o.d -o CMakeFiles/common.dir/speculative.cpp.o -c /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/common/speculative.cpp
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[100%] Linking CXX static library libcommon.a
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/common && /usr/bin/cmake -P CMakeFiles/common.dir/cmake_clean_target.cmake
cd /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/common && /usr/bin/cmake -E cmake_link_script CMakeFiles/common.dir/link.txt --verbose=1
bfd plugin: LLVM gold plugin has failed to create LTO module: Invalid record
bfd plugin: LLVM gold plugin has failed to create LTO module: Invalid record
bfd plugin: LLVM gold plugin has failed to create LTO module: Invalid record
bfd plugin: LLVM gold plugin has failed to create LTO module: Invalid record
bfd plugin: LLVM gold plugin has failed to create LTO module: Invalid record
bfd plugin: LLVM gold plugin has failed to create LTO module: Invalid record
bfd plugin: LLVM gold plugin has failed to create LTO module: Invalid record
bfd plugin: LLVM gold plugin has failed to create LTO module: Invalid record
bfd plugin: LLVM gold plugin has failed to create LTO module: Invalid record
bfd plugin: LLVM gold plugin has failed to create LTO module: Invalid record
bfd plugin: LLVM gold plugin has failed to create LTO module: Invalid record
bfd plugin: LLVM gold plugin has failed to create LTO module: Invalid record
bfd plugin: LLVM gold plugin has failed to create LTO module: Invalid record
bfd plugin: LLVM gold plugin has failed to create LTO module: Invalid record
bfd plugin: LLVM gold plugin has failed to create LTO module: Invalid record
bfd plugin: LLVM gold plugin has failed to create LTO module: Invalid record
bfd plugin: LLVM gold plugin has failed to create LTO module: Invalid record
bfd plugin: LLVM gold plugin has failed to create LTO module: Invalid record
/usr/bin/ar qc libcommon.a CMakeFiles/common.dir/arg.cpp.o CMakeFiles/common.dir/common.cpp.o CMakeFiles/common.dir/console.cpp.o "CMakeFiles/common.dir/json-schema-to-grammar.cpp.o" CMakeFiles/common.dir/log.cpp.o "CMakeFiles/common.dir/ngram-cache.cpp.o" CMakeFiles/common.dir/sampling.cpp.o CMakeFiles/common.dir/speculative.cpp.o "CMakeFiles/build_info.dir/build-info.cpp.o"
/usr/bin/ranlib libcommon.a
gmake[2]: Leaving directory '/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build'
[100%] Built target common
gmake[1]: Leaving directory '/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build'
/usr/bin/cmake -E cmake_progress_start /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build/CMakeFiles 0

+ RPM_EC=0
++ jobs -p
+ exit 0
Executing(%install): /bin/sh -e /var/tmp/rpm-tmp.nOhqXG
+ umask 022
+ cd /builddir/build/BUILD/llama-cpp-b4580-build
+ '[' /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT '!=' / ']'
+ rm -rf /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT
++ dirname /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT
+ mkdir -p /builddir/build/BUILD/llama-cpp-b4580-build
+ mkdir /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT
+ CFLAGS='-O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -fstack-protector-strong   -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer '
+ export CFLAGS
+ CXXFLAGS='-O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -Xarch_host -fstack-protector-strong -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -Xarch_host -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer'
+ export CXXFLAGS
+ FFLAGS='-O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -fstack-protector-strong   -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -I/usr/lib64/gfortran/modules '
+ export FFLAGS
+ FCFLAGS='-O2 -flto=thin -fexceptions -g -grecord-gcc-switches -pipe -Wall -Wp,-U_FORTIFY_SOURCE,-D_FORTIFY_SOURCE=3 -Wp,-D_GLIBCXX_ASSERTIONS --config /usr/lib/rpm/redhat/redhat-hardened-clang.cfg -fstack-protector-strong   -m64 -march=x86-64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -I/usr/lib64/gfortran/modules '
+ export FCFLAGS
+ VALAFLAGS=-g
+ export VALAFLAGS
+ RUSTFLAGS='-Copt-level=3 -Cdebuginfo=2 -Ccodegen-units=1 -Cstrip=none -Cforce-frame-pointers=yes -Clink-arg=-specs=/usr/lib/rpm/redhat/redhat-package-notes --cap-lints=warn'
+ export RUSTFLAGS
+ LDFLAGS='-Wl,-z,relro -Wl,--as-needed  -Wl,-z,pack-relative-relocs -Wl,-z,now -Wl,-z,now   -Wl,--build-id=sha1 -specs=/usr/lib/rpm/redhat/redhat-package-notes '
+ export LDFLAGS
+ LT_SYS_LIBRARY_PATH=/usr/lib64:
+ export LT_SYS_LIBRARY_PATH
+ CC=hipcc
+ export CC
+ CXX=hipcc
+ export CXX
+ cd llama.cpp-b4580
+ DESTDIR=/builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT
+ /usr/bin/cmake --install redhat-linux-build
-- Install configuration: "Release"
-- Installing: /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/lib64/libggml-cpu.so.b4580
-- Installing: /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/lib64/libggml-cpu.so
-- Installing: /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/lib64/libggml-hip.so.b4580
-- Installing: /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/lib64/libggml-hip.so
-- Installing: /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/lib64/libggml.so.b4580
-- Installing: /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/lib64/libggml.so
-- Installing: /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/include/ggml.h
-- Installing: /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/include/ggml-cpu.h
-- Installing: /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/include/ggml-alloc.h
-- Installing: /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/include/ggml-backend.h
-- Installing: /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/include/ggml-blas.h
-- Installing: /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/include/ggml-cann.h
-- Installing: /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/include/ggml-cuda.h
-- Installing: /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/include/ggml-kompute.h
-- Installing: /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/include/ggml-opt.h
-- Installing: /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/include/ggml-metal.h
-- Installing: /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/include/ggml-rpc.h
-- Installing: /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/include/ggml-sycl.h
-- Installing: /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/include/ggml-vulkan.h
-- Installing: /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/include/gguf.h
-- Installing: /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/lib64/libggml-base.so.b4580
-- Installing: /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/lib64/libggml-base.so
-- Installing: /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/lib64/cmake/ggml/ggml-config.cmake
-- Installing: /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/lib64/cmake/ggml/ggml-version.cmake
-- Installing: /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/lib64/libllama.so.b4580
-- Installing: /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/lib64/libllama.so
-- Installing: /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/include/llama.h
-- Installing: /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/include/llama-cpp.h
-- Installing: /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/lib64/cmake/llama/llama-config.cmake
-- Installing: /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/lib64/cmake/llama/llama-version.cmake
-- Installing: /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/bin/convert_hf_to_gguf.py
-- Installing: /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/lib/pkgconfig/llama.pc
+ rm -rf '/builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/lib64/libggml_shared.*'
+ rm /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/bin/convert_hf_to_gguf.py
+ /usr/bin/find-debuginfo -j2 --strict-build-id -m -i --build-id-seed b4580-2.fc43 --unique-debug-suffix -b4580-2.fc43.x86_64 --unique-debug-src-base llama-cpp-b4580-2.fc43.x86_64 --run-dwz --dwz-low-mem-die-limit 10000000 --dwz-max-die-limit 110000000 -S debugsourcefiles.list /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580
find-debuginfo: starting
Extracting debug info from 5 files
DWARF-compressing 5 files
dwz: ./usr/lib64/libggml-base.so.b4580-b4580-2.fc43.x86_64.debug: Unknown debugging section .debug_str_offsets
dwz: ./usr/lib64/libggml-cpu.so.b4580-b4580-2.fc43.x86_64.debug: Unknown debugging section .debug_str_offsets
dwz: ./usr/lib64/libggml-hip.so.b4580-b4580-2.fc43.x86_64.debug: Unknown debugging section .debug_str_offsets
dwz: ./usr/lib64/libggml.so.b4580-b4580-2.fc43.x86_64.debug: Unknown debugging section .debug_str_offsets
dwz: ./usr/lib64/libllama.so.b4580-b4580-2.fc43.x86_64.debug: Unknown debugging section .debug_str_offsets
dwz: Too few files for multifile optimization
sepdebugcrcfix: Updated 0 CRC32s, 5 CRC32s did match.
Creating .debug symlinks for symlinks to ELF files
Copying sources found by 'debugedit -l' to /usr/src/debug/llama-cpp-b4580-2.fc43.x86_64
find-debuginfo: done
+ /usr/lib/rpm/check-buildroot
+ /usr/lib/rpm/redhat/brp-ldconfig
+ /usr/lib/rpm/brp-compress
+ /usr/lib/rpm/redhat/brp-strip-lto /usr/bin/strip
+ /usr/lib/rpm/brp-strip-static-archive /usr/bin/strip
+ /usr/lib/rpm/check-rpaths
+ /usr/lib/rpm/redhat/brp-mangle-shebangs
+ /usr/lib/rpm/brp-remove-la-files
+ env /usr/lib/rpm/redhat/brp-python-bytecompile '' 1 0 -j2
+ /usr/lib/rpm/redhat/brp-python-hardlink
+ /usr/bin/add-determinism --brp -j2 /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT
Scanned 32 directories and 208 files,
               processed 0 inodes,
               0 modified (0 replaced + 0 rewritten),
               0 unsupported format, 0 errors
Reading /builddir/build/BUILD/llama-cpp-b4580-build/SPECPARTS/rpm-debuginfo.specpart
Processing files: llama-cpp-b4580-2.fc43.x86_64
Executing(%license): /bin/sh -e /var/tmp/rpm-tmp.648uEM
+ umask 022
+ cd /builddir/build/BUILD/llama-cpp-b4580-build
+ cd llama.cpp-b4580
+ LICENSEDIR=/builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/share/licenses/llama-cpp
+ export LC_ALL=C.UTF-8
+ LC_ALL=C.UTF-8
+ export LICENSEDIR
+ /usr/bin/mkdir -p /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/share/licenses/llama-cpp
+ cp -pr /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/LICENSE /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/share/licenses/llama-cpp
+ RPM_EC=0
++ jobs -p
+ exit 0
Provides: libggml-base.so.b4580()(64bit) libggml-cpu.so.b4580()(64bit) libggml-hip.so.b4580()(64bit) libggml.so.b4580()(64bit) libllama.so.b4580()(64bit) llama-cpp = b4580-2.fc43 llama-cpp(x86-64) = b4580-2.fc43
Requires(rpmlib): rpmlib(CompressedFileNames) <= 3.0.4-1 rpmlib(FileDigests) <= 4.6.0-1 rpmlib(PayloadFilesHavePrefix) <= 4.0-1
Requires: ld-linux-x86-64.so.2()(64bit) ld-linux-x86-64.so.2(GLIBC_2.3)(64bit) libamdhip64.so.6()(64bit) libamdhip64.so.6(hip_4.2)(64bit) libamdhip64.so.6(hip_6.0)(64bit) libc.so.6()(64bit) libc.so.6(GLIBC_2.14)(64bit) libc.so.6(GLIBC_2.17)(64bit) libc.so.6(GLIBC_2.2.5)(64bit) libc.so.6(GLIBC_2.29)(64bit) libc.so.6(GLIBC_2.3.2)(64bit) libc.so.6(GLIBC_2.3.4)(64bit) libc.so.6(GLIBC_2.32)(64bit) libc.so.6(GLIBC_2.33)(64bit) libc.so.6(GLIBC_2.34)(64bit) libc.so.6(GLIBC_2.38)(64bit) libc.so.6(GLIBC_2.4)(64bit) libc.so.6(GLIBC_2.7)(64bit) libc.so.6(GLIBC_ABI_DT_RELR)(64bit) libgcc_s.so.1()(64bit) libgcc_s.so.1(GCC_3.0)(64bit) libggml-base.so.b4580()(64bit) libggml-cpu.so.b4580()(64bit) libggml-hip.so.b4580()(64bit) libggml.so.b4580()(64bit) libhipblas.so.2()(64bit) libm.so.6()(64bit) libm.so.6(GLIBC_2.2.5)(64bit) libm.so.6(GLIBC_2.27)(64bit) libm.so.6(GLIBC_2.29)(64bit) librocblas.so.4()(64bit) libstdc++.so.6()(64bit) libstdc++.so.6(CXXABI_1.3)(64bit) libstdc++.so.6(CXXABI_1.3.11)(64bit) libstdc++.so.6(CXXABI_1.3.13)(64bit) libstdc++.so.6(CXXABI_1.3.2)(64bit) libstdc++.so.6(CXXABI_1.3.3)(64bit) libstdc++.so.6(CXXABI_1.3.5)(64bit) libstdc++.so.6(GLIBCXX_3.4)(64bit) libstdc++.so.6(GLIBCXX_3.4.11)(64bit) libstdc++.so.6(GLIBCXX_3.4.14)(64bit) libstdc++.so.6(GLIBCXX_3.4.15)(64bit) libstdc++.so.6(GLIBCXX_3.4.17)(64bit) libstdc++.so.6(GLIBCXX_3.4.18)(64bit) libstdc++.so.6(GLIBCXX_3.4.19)(64bit) libstdc++.so.6(GLIBCXX_3.4.20)(64bit) libstdc++.so.6(GLIBCXX_3.4.21)(64bit) libstdc++.so.6(GLIBCXX_3.4.22)(64bit) libstdc++.so.6(GLIBCXX_3.4.25)(64bit) libstdc++.so.6(GLIBCXX_3.4.26)(64bit) libstdc++.so.6(GLIBCXX_3.4.29)(64bit) libstdc++.so.6(GLIBCXX_3.4.30)(64bit) libstdc++.so.6(GLIBCXX_3.4.9)(64bit)
Recommends: numactl
Processing files: llama-cpp-devel-b4580-2.fc43.x86_64
Executing(%doc): /bin/sh -e /var/tmp/rpm-tmp.oHGhJW
+ umask 022
+ cd /builddir/build/BUILD/llama-cpp-b4580-build
+ cd llama.cpp-b4580
+ DOCDIR=/builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/share/doc/llama-cpp-devel
+ export LC_ALL=C.UTF-8
+ LC_ALL=C.UTF-8
+ export DOCDIR
+ /usr/bin/mkdir -p /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/share/doc/llama-cpp-devel
+ cp -pr /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/README.md /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT/usr/share/doc/llama-cpp-devel
+ RPM_EC=0
++ jobs -p
+ exit 0
Provides: cmake(ggml) cmake(llama) llama-cpp-devel = b4580-2.fc43 llama-cpp-devel(x86-64) = b4580-2.fc43
Requires(rpmlib): rpmlib(CompressedFileNames) <= 3.0.4-1 rpmlib(FileDigests) <= 4.6.0-1 rpmlib(PayloadFilesHavePrefix) <= 4.0-1
Requires: cmake-filesystem(x86-64) libggml-base.so.b4580()(64bit) libggml-cpu.so.b4580()(64bit) libggml-hip.so.b4580()(64bit) libggml.so.b4580()(64bit) libllama.so.b4580()(64bit)
Processing files: llama-cpp-debugsource-b4580-2.fc43.x86_64
Provides: llama-cpp-debugsource = b4580-2.fc43 llama-cpp-debugsource(x86-64) = b4580-2.fc43
Requires(rpmlib): rpmlib(CompressedFileNames) <= 3.0.4-1 rpmlib(FileDigests) <= 4.6.0-1 rpmlib(PayloadFilesHavePrefix) <= 4.0-1
Processing files: llama-cpp-debuginfo-b4580-2.fc43.x86_64
Provides: debuginfo(build-id) = 53c492c386d774e4cd07829f7cef45012f70b492 debuginfo(build-id) = 8781f2c64e74e163370ea72b80b9448203c8c030 debuginfo(build-id) = ae429299d59c82156c7a17ed17059e5010a720b2 debuginfo(build-id) = ec4948c9e76adb1935305a97cb2f217ddf95c0cf debuginfo(build-id) = ed00045dde200d19686c6e5b58d3ea5a78c46783 libggml-base.so.b4580-b4580-2.fc43.x86_64.debug()(64bit) libggml-cpu.so.b4580-b4580-2.fc43.x86_64.debug()(64bit) libggml-hip.so.b4580-b4580-2.fc43.x86_64.debug()(64bit) libggml.so.b4580-b4580-2.fc43.x86_64.debug()(64bit) libllama.so.b4580-b4580-2.fc43.x86_64.debug()(64bit) llama-cpp-debuginfo = b4580-2.fc43 llama-cpp-debuginfo(x86-64) = b4580-2.fc43
Requires(rpmlib): rpmlib(CompressedFileNames) <= 3.0.4-1 rpmlib(FileDigests) <= 4.6.0-1 rpmlib(PayloadFilesHavePrefix) <= 4.0-1
Recommends: llama-cpp-debugsource(x86-64) = b4580-2.fc43
Checking for unpackaged file(s): /usr/lib/rpm/check-files /builddir/build/BUILD/llama-cpp-b4580-build/BUILDROOT
Wrote: /builddir/build/RPMS/llama-cpp-debuginfo-b4580-2.fc43.x86_64.rpm
Wrote: /builddir/build/RPMS/llama-cpp-debugsource-b4580-2.fc43.x86_64.rpm
Wrote: /builddir/build/RPMS/llama-cpp-devel-b4580-2.fc43.x86_64.rpm
Wrote: /builddir/build/RPMS/llama-cpp-b4580-2.fc43.x86_64.rpm
Executing(rmbuild): /bin/sh -e /var/tmp/rpm-tmp.PXlbMU
+ umask 022
+ cd /builddir/build/BUILD/llama-cpp-b4580-build
+ test -d /builddir/build/BUILD/llama-cpp-b4580-build
+ /usr/bin/chmod -Rf a+rX,u+w,g-w,o-w /builddir/build/BUILD/llama-cpp-b4580-build
+ rm -rf /builddir/build/BUILD/llama-cpp-b4580-build
+ RPM_EC=0
++ jobs -p
+ exit 0
Finish: rpmbuild llama-cpp-b4580-2.fc43.src.rpm
Finish: build phase for llama-cpp-b4580-2.fc43.src.rpm
INFO: chroot_scan: 1 files copied to /var/lib/copr-rpmbuild/results/chroot_scan
INFO: /var/lib/mock/fedora-rawhide-x86_64-1742774257.650632/root/var/log/dnf5.log
INFO: chroot_scan: creating tarball /var/lib/copr-rpmbuild/results/chroot_scan.tar.gz
/bin/tar: Removing leading `/' from member names
INFO: Done(/var/lib/copr-rpmbuild/results/llama-cpp-b4580-2.fc43.src.rpm) Config(child) 157 minutes 32 seconds
INFO: Results and/or logs in: /var/lib/copr-rpmbuild/results
INFO: Cleaning up build root ('cleanup_on_success=True')
Start: clean chroot
INFO: unmounting tmpfs.
Finish: clean chroot
Finish: run
Running RPMResults tool
Package info:
{
    "packages": [
        {
            "name": "llama-cpp",
            "epoch": null,
            "version": "b4580",
            "release": "2.fc43",
            "arch": "x86_64"
        },
        {
            "name": "llama-cpp-debuginfo",
            "epoch": null,
            "version": "b4580",
            "release": "2.fc43",
            "arch": "x86_64"
        },
        {
            "name": "llama-cpp-devel",
            "epoch": null,
            "version": "b4580",
            "release": "2.fc43",
            "arch": "x86_64"
        },
        {
            "name": "llama-cpp",
            "epoch": null,
            "version": "b4580",
            "release": "2.fc43",
            "arch": "src"
        },
        {
            "name": "llama-cpp-debugsource",
            "epoch": null,
            "version": "b4580",
            "release": "2.fc43",
            "arch": "x86_64"
        }
    ]
}
RPMResults finished